diff --git a/test/Transforms/SLPVectorizer/X86/arith-abs.ll b/test/Transforms/SLPVectorizer/X86/arith-abs.ll
new file mode 100644
index 00000000000..a5d151e6107
--- /dev/null
+++ b/test/Transforms/SLPVectorizer/X86/arith-abs.ll
@@ -0,0 +1,582 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -mtriple=x86_64-unknown -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,SSE
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=slm -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,SLM
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=corei7-avx -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=knl -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512F
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=-prefer-256-bit -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512BW
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=+prefer-256-bit -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX256BW
+
+@a64 = common global [8 x i64] zeroinitializer, align 64
+@b64 = common global [8 x i64] zeroinitializer, align 64
+@c64 = common global [8 x i64] zeroinitializer, align 64
+@a32 = common global [16 x i32] zeroinitializer, align 64
+@b32 = common global [16 x i32] zeroinitializer, align 64
+@c32 = common global [16 x i32] zeroinitializer, align 64
+@a16 = common global [32 x i16] zeroinitializer, align 64
+@b16 = common global [32 x i16] zeroinitializer, align 64
+@c16 = common global [32 x i16] zeroinitializer, align 64
+@a8  = common global [64 x i8] zeroinitializer, align 64
+@b8  = common global [64 x i8] zeroinitializer, align 64
+@c8  = common global [64 x i8] zeroinitializer, align 64
+
+declare i64 @llvm.abs.i64(i64, i1)
+declare i32 @llvm.abs.i32(i32, i1)
+declare i16 @llvm.abs.i16(i16, i1)
+declare i8  @llvm.abs.i8 (i8, i1)
+
+define void @abs_v8i64() {
+; SSE-LABEL: @abs_v8i64(
+; SSE-NEXT:    [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @a64 to <2 x i64>*), align 8
+; SSE-NEXT:    [[TMP2:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2) to <2 x i64>*), align 8
+; SSE-NEXT:    [[TMP3:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <2 x i64>*), align 8
+; SSE-NEXT:    [[TMP4:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6) to <2 x i64>*), align 8
+; SSE-NEXT:    [[TMP5:%.*]] = call <2 x i64> @llvm.abs.v2i64(<2 x i64> [[TMP1]], i1 false)
+; SSE-NEXT:    [[TMP6:%.*]] = call <2 x i64> @llvm.abs.v2i64(<2 x i64> [[TMP2]], i1 false)
+; SSE-NEXT:    [[TMP7:%.*]] = call <2 x i64> @llvm.abs.v2i64(<2 x i64> [[TMP3]], i1 false)
+; SSE-NEXT:    [[TMP8:%.*]] = call <2 x i64> @llvm.abs.v2i64(<2 x i64> [[TMP4]], i1 false)
+; SSE-NEXT:    store <2 x i64> [[TMP5]], <2 x i64>* bitcast ([8 x i64]* @c64 to <2 x i64>*), align 8
+; SSE-NEXT:    store <2 x i64> [[TMP6]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2) to <2 x i64>*), align 8
+; SSE-NEXT:    store <2 x i64> [[TMP7]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <2 x i64>*), align 8
+; SSE-NEXT:    store <2 x i64> [[TMP8]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6) to <2 x i64>*), align 8
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @abs_v8i64(
+; SLM-NEXT:    [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @a64 to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP2:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP3:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP4:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP5:%.*]] = call <2 x i64> @llvm.abs.v2i64(<2 x i64> [[TMP1]], i1 false)
+; SLM-NEXT:    [[TMP6:%.*]] = call <2 x i64> @llvm.abs.v2i64(<2 x i64> [[TMP2]], i1 false)
+; SLM-NEXT:    [[TMP7:%.*]] = call <2 x i64> @llvm.abs.v2i64(<2 x i64> [[TMP3]], i1 false)
+; SLM-NEXT:    [[TMP8:%.*]] = call <2 x i64> @llvm.abs.v2i64(<2 x i64> [[TMP4]], i1 false)
+; SLM-NEXT:    store <2 x i64> [[TMP5]], <2 x i64>* bitcast ([8 x i64]* @c64 to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP6]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP7]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP8]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @abs_v8i64(
+; AVX-NEXT:    [[TMP1:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @a64 to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP2:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP3:%.*]] = call <4 x i64> @llvm.abs.v4i64(<4 x i64> [[TMP1]], i1 false)
+; AVX-NEXT:    [[TMP4:%.*]] = call <4 x i64> @llvm.abs.v4i64(<4 x i64> [[TMP2]], i1 false)
+; AVX-NEXT:    store <4 x i64> [[TMP3]], <4 x i64>* bitcast ([8 x i64]* @c64 to <4 x i64>*), align 8
+; AVX-NEXT:    store <4 x i64> [[TMP4]], <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @abs_v8i64(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <8 x i64>, <8 x i64>* bitcast ([8 x i64]* @a64 to <8 x i64>*), align 8
+; AVX512-NEXT:    [[TMP2:%.*]] = call <8 x i64> @llvm.abs.v8i64(<8 x i64> [[TMP1]], i1 false)
+; AVX512-NEXT:    store <8 x i64> [[TMP2]], <8 x i64>* bitcast ([8 x i64]* @c64 to <8 x i64>*), align 8
+; AVX512-NEXT:    ret void
+;
+  %a0 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8
+  %a1 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8
+  %a2 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8
+  %a3 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8
+  %a4 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8
+  %a5 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8
+  %a6 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8
+  %a7 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8
+  %r0 = call i64 @llvm.abs.i64(i64 %a0, i1 false)
+  %r1 = call i64 @llvm.abs.i64(i64 %a1, i1 false)
+  %r2 = call i64 @llvm.abs.i64(i64 %a2, i1 false)
+  %r3 = call i64 @llvm.abs.i64(i64 %a3, i1 false)
+  %r4 = call i64 @llvm.abs.i64(i64 %a4, i1 false)
+  %r5 = call i64 @llvm.abs.i64(i64 %a5, i1 false)
+  %r6 = call i64 @llvm.abs.i64(i64 %a6, i1 false)
+  %r7 = call i64 @llvm.abs.i64(i64 %a7, i1 false)
+  store i64 %r0, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8
+  store i64 %r1, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8
+  store i64 %r2, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8
+  store i64 %r3, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8
+  store i64 %r4, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8
+  store i64 %r5, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8
+  store i64 %r6, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8
+  store i64 %r7, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8
+  ret void
+}
+
+define void @abs_v16i32() {
+; SSE-LABEL: @abs_v16i32(
+; SSE-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP5:%.*]] = call <4 x i32> @llvm.abs.v4i32(<4 x i32> [[TMP1]], i1 false)
+; SSE-NEXT:    [[TMP6:%.*]] = call <4 x i32> @llvm.abs.v4i32(<4 x i32> [[TMP2]], i1 false)
+; SSE-NEXT:    [[TMP7:%.*]] = call <4 x i32> @llvm.abs.v4i32(<4 x i32> [[TMP3]], i1 false)
+; SSE-NEXT:    [[TMP8:%.*]] = call <4 x i32> @llvm.abs.v4i32(<4 x i32> [[TMP4]], i1 false)
+; SSE-NEXT:    store <4 x i32> [[TMP5]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP6]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP7]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP8]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @abs_v16i32(
+; SLM-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP5:%.*]] = call <4 x i32> @llvm.abs.v4i32(<4 x i32> [[TMP1]], i1 false)
+; SLM-NEXT:    [[TMP6:%.*]] = call <4 x i32> @llvm.abs.v4i32(<4 x i32> [[TMP2]], i1 false)
+; SLM-NEXT:    [[TMP7:%.*]] = call <4 x i32> @llvm.abs.v4i32(<4 x i32> [[TMP3]], i1 false)
+; SLM-NEXT:    [[TMP8:%.*]] = call <4 x i32> @llvm.abs.v4i32(<4 x i32> [[TMP4]], i1 false)
+; SLM-NEXT:    store <4 x i32> [[TMP5]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP6]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP7]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP8]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @abs_v16i32(
+; AVX-NEXT:    [[TMP1:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([16 x i32]* @a32 to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP3:%.*]] = call <8 x i32> @llvm.abs.v8i32(<8 x i32> [[TMP1]], i1 false)
+; AVX-NEXT:    [[TMP4:%.*]] = call <8 x i32> @llvm.abs.v8i32(<8 x i32> [[TMP2]], i1 false)
+; AVX-NEXT:    store <8 x i32> [[TMP3]], <8 x i32>* bitcast ([16 x i32]* @c32 to <8 x i32>*), align 4
+; AVX-NEXT:    store <8 x i32> [[TMP4]], <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @abs_v16i32(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <16 x i32>, <16 x i32>* bitcast ([16 x i32]* @a32 to <16 x i32>*), align 4
+; AVX512-NEXT:    [[TMP2:%.*]] = call <16 x i32> @llvm.abs.v16i32(<16 x i32> [[TMP1]], i1 false)
+; AVX512-NEXT:    store <16 x i32> [[TMP2]], <16 x i32>* bitcast ([16 x i32]* @c32 to <16 x i32>*), align 4
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 0 ), align 4
+  %a1  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 1 ), align 4
+  %a2  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 2 ), align 4
+  %a3  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 3 ), align 4
+  %a4  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4 ), align 4
+  %a5  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 5 ), align 4
+  %a6  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 6 ), align 4
+  %a7  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 7 ), align 4
+  %a8  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8 ), align 4
+  %a9  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 9 ), align 4
+  %a10 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 10), align 4
+  %a11 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 11), align 4
+  %a12 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12), align 4
+  %a13 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 13), align 4
+  %a14 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 14), align 4
+  %a15 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 15), align 4
+  %r0  = call i32 @llvm.abs.i32(i32 %a0, i1 false)
+  %r1  = call i32 @llvm.abs.i32(i32 %a1, i1 false)
+  %r2  = call i32 @llvm.abs.i32(i32 %a2, i1 false)
+  %r3  = call i32 @llvm.abs.i32(i32 %a3, i1 false)
+  %r4  = call i32 @llvm.abs.i32(i32 %a4, i1 false)
+  %r5  = call i32 @llvm.abs.i32(i32 %a5, i1 false)
+  %r6  = call i32 @llvm.abs.i32(i32 %a6, i1 false)
+  %r7  = call i32 @llvm.abs.i32(i32 %a7, i1 false)
+  %r8  = call i32 @llvm.abs.i32(i32 %a8, i1 false)
+  %r9  = call i32 @llvm.abs.i32(i32 %a9, i1 false)
+  %r10 = call i32 @llvm.abs.i32(i32 %a10, i1 false)
+  %r11 = call i32 @llvm.abs.i32(i32 %a11, i1 false)
+  %r12 = call i32 @llvm.abs.i32(i32 %a12, i1 false)
+  %r13 = call i32 @llvm.abs.i32(i32 %a13, i1 false)
+  %r14 = call i32 @llvm.abs.i32(i32 %a14, i1 false)
+  %r15 = call i32 @llvm.abs.i32(i32 %a15, i1 false)
+  store i32 %r0 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 0 ), align 4
+  store i32 %r1 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 1 ), align 4
+  store i32 %r2 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 2 ), align 4
+  store i32 %r3 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 3 ), align 4
+  store i32 %r4 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4 ), align 4
+  store i32 %r5 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 5 ), align 4
+  store i32 %r6 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 6 ), align 4
+  store i32 %r7 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 7 ), align 4
+  store i32 %r8 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8 ), align 4
+  store i32 %r9 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 9 ), align 4
+  store i32 %r10, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 10), align 4
+  store i32 %r11, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 11), align 4
+  store i32 %r12, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12), align 4
+  store i32 %r13, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 13), align 4
+  store i32 %r14, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 14), align 4
+  store i32 %r15, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 15), align 4
+  ret void
+}
+
+define void @abs_v32i16() {
+; SSE-LABEL: @abs_v32i16(
+; SSE-NEXT:    [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @a16 to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP3:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP5:%.*]] = call <8 x i16> @llvm.abs.v8i16(<8 x i16> [[TMP1]], i1 false)
+; SSE-NEXT:    [[TMP6:%.*]] = call <8 x i16> @llvm.abs.v8i16(<8 x i16> [[TMP2]], i1 false)
+; SSE-NEXT:    [[TMP7:%.*]] = call <8 x i16> @llvm.abs.v8i16(<8 x i16> [[TMP3]], i1 false)
+; SSE-NEXT:    [[TMP8:%.*]] = call <8 x i16> @llvm.abs.v8i16(<8 x i16> [[TMP4]], i1 false)
+; SSE-NEXT:    store <8 x i16> [[TMP5]], <8 x i16>* bitcast ([32 x i16]* @c16 to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP6]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP7]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP8]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @abs_v32i16(
+; SLM-NEXT:    [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @a16 to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP3:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP5:%.*]] = call <8 x i16> @llvm.abs.v8i16(<8 x i16> [[TMP1]], i1 false)
+; SLM-NEXT:    [[TMP6:%.*]] = call <8 x i16> @llvm.abs.v8i16(<8 x i16> [[TMP2]], i1 false)
+; SLM-NEXT:    [[TMP7:%.*]] = call <8 x i16> @llvm.abs.v8i16(<8 x i16> [[TMP3]], i1 false)
+; SLM-NEXT:    [[TMP8:%.*]] = call <8 x i16> @llvm.abs.v8i16(<8 x i16> [[TMP4]], i1 false)
+; SLM-NEXT:    store <8 x i16> [[TMP5]], <8 x i16>* bitcast ([32 x i16]* @c16 to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP6]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP7]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP8]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @abs_v32i16(
+; AVX-NEXT:    [[TMP1:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([32 x i16]* @a16 to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP2:%.*]] = load <16 x i16>, <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP3:%.*]] = call <16 x i16> @llvm.abs.v16i16(<16 x i16> [[TMP1]], i1 false)
+; AVX-NEXT:    [[TMP4:%.*]] = call <16 x i16> @llvm.abs.v16i16(<16 x i16> [[TMP2]], i1 false)
+; AVX-NEXT:    store <16 x i16> [[TMP3]], <16 x i16>* bitcast ([32 x i16]* @c16 to <16 x i16>*), align 2
+; AVX-NEXT:    store <16 x i16> [[TMP4]], <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @abs_v32i16(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <32 x i16>, <32 x i16>* bitcast ([32 x i16]* @a16 to <32 x i16>*), align 2
+; AVX512-NEXT:    [[TMP2:%.*]] = call <32 x i16> @llvm.abs.v32i16(<32 x i16> [[TMP1]], i1 false)
+; AVX512-NEXT:    store <32 x i16> [[TMP2]], <32 x i16>* bitcast ([32 x i16]* @c16 to <32 x i16>*), align 2
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 0 ), align 2
+  %a1  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 1 ), align 2
+  %a2  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 2 ), align 2
+  %a3  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 3 ), align 2
+  %a4  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 4 ), align 2
+  %a5  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 5 ), align 2
+  %a6  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 6 ), align 2
+  %a7  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 7 ), align 2
+  %a8  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8 ), align 2
+  %a9  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 9 ), align 2
+  %a10 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 10), align 2
+  %a11 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 11), align 2
+  %a12 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 12), align 2
+  %a13 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 13), align 2
+  %a14 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 14), align 2
+  %a15 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 15), align 2
+  %a16 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16), align 2
+  %a17 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 17), align 2
+  %a18 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 18), align 2
+  %a19 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 19), align 2
+  %a20 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 20), align 2
+  %a21 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 21), align 2
+  %a22 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 22), align 2
+  %a23 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 23), align 2
+  %a24 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24), align 2
+  %a25 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 25), align 2
+  %a26 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 26), align 2
+  %a27 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 27), align 2
+  %a28 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 28), align 2
+  %a29 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 29), align 2
+  %a30 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 30), align 2
+  %a31 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 31), align 2
+  %r0  = call i16 @llvm.abs.i16(i16 %a0, i1 false)
+  %r1  = call i16 @llvm.abs.i16(i16 %a1, i1 false)
+  %r2  = call i16 @llvm.abs.i16(i16 %a2, i1 false)
+  %r3  = call i16 @llvm.abs.i16(i16 %a3, i1 false)
+  %r4  = call i16 @llvm.abs.i16(i16 %a4, i1 false)
+  %r5  = call i16 @llvm.abs.i16(i16 %a5, i1 false)
+  %r6  = call i16 @llvm.abs.i16(i16 %a6, i1 false)
+  %r7  = call i16 @llvm.abs.i16(i16 %a7, i1 false)
+  %r8  = call i16 @llvm.abs.i16(i16 %a8, i1 false)
+  %r9  = call i16 @llvm.abs.i16(i16 %a9, i1 false)
+  %r10 = call i16 @llvm.abs.i16(i16 %a10, i1 false)
+  %r11 = call i16 @llvm.abs.i16(i16 %a11, i1 false)
+  %r12 = call i16 @llvm.abs.i16(i16 %a12, i1 false)
+  %r13 = call i16 @llvm.abs.i16(i16 %a13, i1 false)
+  %r14 = call i16 @llvm.abs.i16(i16 %a14, i1 false)
+  %r15 = call i16 @llvm.abs.i16(i16 %a15, i1 false)
+  %r16 = call i16 @llvm.abs.i16(i16 %a16, i1 false)
+  %r17 = call i16 @llvm.abs.i16(i16 %a17, i1 false)
+  %r18 = call i16 @llvm.abs.i16(i16 %a18, i1 false)
+  %r19 = call i16 @llvm.abs.i16(i16 %a19, i1 false)
+  %r20 = call i16 @llvm.abs.i16(i16 %a20, i1 false)
+  %r21 = call i16 @llvm.abs.i16(i16 %a21, i1 false)
+  %r22 = call i16 @llvm.abs.i16(i16 %a22, i1 false)
+  %r23 = call i16 @llvm.abs.i16(i16 %a23, i1 false)
+  %r24 = call i16 @llvm.abs.i16(i16 %a24, i1 false)
+  %r25 = call i16 @llvm.abs.i16(i16 %a25, i1 false)
+  %r26 = call i16 @llvm.abs.i16(i16 %a26, i1 false)
+  %r27 = call i16 @llvm.abs.i16(i16 %a27, i1 false)
+  %r28 = call i16 @llvm.abs.i16(i16 %a28, i1 false)
+  %r29 = call i16 @llvm.abs.i16(i16 %a29, i1 false)
+  %r30 = call i16 @llvm.abs.i16(i16 %a30, i1 false)
+  %r31 = call i16 @llvm.abs.i16(i16 %a31, i1 false)
+  store i16 %r0 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 0 ), align 2
+  store i16 %r1 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 1 ), align 2
+  store i16 %r2 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 2 ), align 2
+  store i16 %r3 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 3 ), align 2
+  store i16 %r4 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 4 ), align 2
+  store i16 %r5 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 5 ), align 2
+  store i16 %r6 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 6 ), align 2
+  store i16 %r7 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 7 ), align 2
+  store i16 %r8 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8 ), align 2
+  store i16 %r9 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 9 ), align 2
+  store i16 %r10, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 10), align 2
+  store i16 %r11, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 11), align 2
+  store i16 %r12, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 12), align 2
+  store i16 %r13, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 13), align 2
+  store i16 %r14, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 14), align 2
+  store i16 %r15, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 15), align 2
+  store i16 %r16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16), align 2
+  store i16 %r17, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 17), align 2
+  store i16 %r18, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 18), align 2
+  store i16 %r19, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 19), align 2
+  store i16 %r20, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 20), align 2
+  store i16 %r21, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 21), align 2
+  store i16 %r22, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 22), align 2
+  store i16 %r23, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 23), align 2
+  store i16 %r24, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24), align 2
+  store i16 %r25, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 25), align 2
+  store i16 %r26, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 26), align 2
+  store i16 %r27, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 27), align 2
+  store i16 %r28, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 28), align 2
+  store i16 %r29, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 29), align 2
+  store i16 %r30, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 30), align 2
+  store i16 %r31, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 31), align 2
+  ret void
+}
+
+define void @abs_v64i8() {
+; SSE-LABEL: @abs_v64i8(
+; SSE-NEXT:    [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @a8 to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP5:%.*]] = call <16 x i8> @llvm.abs.v16i8(<16 x i8> [[TMP1]], i1 false)
+; SSE-NEXT:    [[TMP6:%.*]] = call <16 x i8> @llvm.abs.v16i8(<16 x i8> [[TMP2]], i1 false)
+; SSE-NEXT:    [[TMP7:%.*]] = call <16 x i8> @llvm.abs.v16i8(<16 x i8> [[TMP3]], i1 false)
+; SSE-NEXT:    [[TMP8:%.*]] = call <16 x i8> @llvm.abs.v16i8(<16 x i8> [[TMP4]], i1 false)
+; SSE-NEXT:    store <16 x i8> [[TMP5]], <16 x i8>* bitcast ([64 x i8]* @c8 to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP6]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP7]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP8]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @abs_v64i8(
+; SLM-NEXT:    [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @a8 to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP5:%.*]] = call <16 x i8> @llvm.abs.v16i8(<16 x i8> [[TMP1]], i1 false)
+; SLM-NEXT:    [[TMP6:%.*]] = call <16 x i8> @llvm.abs.v16i8(<16 x i8> [[TMP2]], i1 false)
+; SLM-NEXT:    [[TMP7:%.*]] = call <16 x i8> @llvm.abs.v16i8(<16 x i8> [[TMP3]], i1 false)
+; SLM-NEXT:    [[TMP8:%.*]] = call <16 x i8> @llvm.abs.v16i8(<16 x i8> [[TMP4]], i1 false)
+; SLM-NEXT:    store <16 x i8> [[TMP5]], <16 x i8>* bitcast ([64 x i8]* @c8 to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP6]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP7]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP8]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @abs_v64i8(
+; AVX-NEXT:    [[TMP1:%.*]] = load <32 x i8>, <32 x i8>* bitcast ([64 x i8]* @a8 to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP2:%.*]] = load <32 x i8>, <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP3:%.*]] = call <32 x i8> @llvm.abs.v32i8(<32 x i8> [[TMP1]], i1 false)
+; AVX-NEXT:    [[TMP4:%.*]] = call <32 x i8> @llvm.abs.v32i8(<32 x i8> [[TMP2]], i1 false)
+; AVX-NEXT:    store <32 x i8> [[TMP3]], <32 x i8>* bitcast ([64 x i8]* @c8 to <32 x i8>*), align 1
+; AVX-NEXT:    store <32 x i8> [[TMP4]], <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @abs_v64i8(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <64 x i8>, <64 x i8>* bitcast ([64 x i8]* @a8 to <64 x i8>*), align 1
+; AVX512-NEXT:    [[TMP2:%.*]] = call <64 x i8> @llvm.abs.v64i8(<64 x i8> [[TMP1]], i1 false)
+; AVX512-NEXT:    store <64 x i8> [[TMP2]], <64 x i8>* bitcast ([64 x i8]* @c8 to <64 x i8>*), align 1
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 0 ), align 1
+  %a1  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 1 ), align 1
+  %a2  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 2 ), align 1
+  %a3  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 3 ), align 1
+  %a4  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 4 ), align 1
+  %a5  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 5 ), align 1
+  %a6  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 6 ), align 1
+  %a7  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 7 ), align 1
+  %a8  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 8 ), align 1
+  %a9  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 9 ), align 1
+  %a10 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 10), align 1
+  %a11 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 11), align 1
+  %a12 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 12), align 1
+  %a13 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 13), align 1
+  %a14 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 14), align 1
+  %a15 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 15), align 1
+  %a16 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16), align 1
+  %a17 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 17), align 1
+  %a18 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 18), align 1
+  %a19 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 19), align 1
+  %a20 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 20), align 1
+  %a21 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 21), align 1
+  %a22 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 22), align 1
+  %a23 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 23), align 1
+  %a24 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 24), align 1
+  %a25 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 25), align 1
+  %a26 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 26), align 1
+  %a27 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 27), align 1
+  %a28 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 28), align 1
+  %a29 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 29), align 1
+  %a30 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 30), align 1
+  %a31 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 31), align 1
+  %a32 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32), align 1
+  %a33 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 33), align 1
+  %a34 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 34), align 1
+  %a35 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 35), align 1
+  %a36 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 36), align 1
+  %a37 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 37), align 1
+  %a38 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 38), align 1
+  %a39 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 39), align 1
+  %a40 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 40), align 1
+  %a41 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 41), align 1
+  %a42 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 42), align 1
+  %a43 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 43), align 1
+  %a44 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 44), align 1
+  %a45 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 45), align 1
+  %a46 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 46), align 1
+  %a47 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 47), align 1
+  %a48 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48), align 1
+  %a49 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 49), align 1
+  %a50 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 50), align 1
+  %a51 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 51), align 1
+  %a52 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 52), align 1
+  %a53 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 53), align 1
+  %a54 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 54), align 1
+  %a55 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 55), align 1
+  %a56 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 56), align 1
+  %a57 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 57), align 1
+  %a58 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 58), align 1
+  %a59 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 59), align 1
+  %a60 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 60), align 1
+  %a61 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 61), align 1
+  %a62 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 62), align 1
+  %a63 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 63), align 1
+  %r0  = call i8 @llvm.abs.i8(i8 %a0, i1 false)
+  %r1  = call i8 @llvm.abs.i8(i8 %a1, i1 false)
+  %r2  = call i8 @llvm.abs.i8(i8 %a2, i1 false)
+  %r3  = call i8 @llvm.abs.i8(i8 %a3, i1 false)
+  %r4  = call i8 @llvm.abs.i8(i8 %a4, i1 false)
+  %r5  = call i8 @llvm.abs.i8(i8 %a5, i1 false)
+  %r6  = call i8 @llvm.abs.i8(i8 %a6, i1 false)
+  %r7  = call i8 @llvm.abs.i8(i8 %a7, i1 false)
+  %r8  = call i8 @llvm.abs.i8(i8 %a8, i1 false)
+  %r9  = call i8 @llvm.abs.i8(i8 %a9, i1 false)
+  %r10 = call i8 @llvm.abs.i8(i8 %a10, i1 false)
+  %r11 = call i8 @llvm.abs.i8(i8 %a11, i1 false)
+  %r12 = call i8 @llvm.abs.i8(i8 %a12, i1 false)
+  %r13 = call i8 @llvm.abs.i8(i8 %a13, i1 false)
+  %r14 = call i8 @llvm.abs.i8(i8 %a14, i1 false)
+  %r15 = call i8 @llvm.abs.i8(i8 %a15, i1 false)
+  %r16 = call i8 @llvm.abs.i8(i8 %a16, i1 false)
+  %r17 = call i8 @llvm.abs.i8(i8 %a17, i1 false)
+  %r18 = call i8 @llvm.abs.i8(i8 %a18, i1 false)
+  %r19 = call i8 @llvm.abs.i8(i8 %a19, i1 false)
+  %r20 = call i8 @llvm.abs.i8(i8 %a20, i1 false)
+  %r21 = call i8 @llvm.abs.i8(i8 %a21, i1 false)
+  %r22 = call i8 @llvm.abs.i8(i8 %a22, i1 false)
+  %r23 = call i8 @llvm.abs.i8(i8 %a23, i1 false)
+  %r24 = call i8 @llvm.abs.i8(i8 %a24, i1 false)
+  %r25 = call i8 @llvm.abs.i8(i8 %a25, i1 false)
+  %r26 = call i8 @llvm.abs.i8(i8 %a26, i1 false)
+  %r27 = call i8 @llvm.abs.i8(i8 %a27, i1 false)
+  %r28 = call i8 @llvm.abs.i8(i8 %a28, i1 false)
+  %r29 = call i8 @llvm.abs.i8(i8 %a29, i1 false)
+  %r30 = call i8 @llvm.abs.i8(i8 %a30, i1 false)
+  %r31 = call i8 @llvm.abs.i8(i8 %a31, i1 false)
+  %r32 = call i8 @llvm.abs.i8(i8 %a32, i1 false)
+  %r33 = call i8 @llvm.abs.i8(i8 %a33, i1 false)
+  %r34 = call i8 @llvm.abs.i8(i8 %a34, i1 false)
+  %r35 = call i8 @llvm.abs.i8(i8 %a35, i1 false)
+  %r36 = call i8 @llvm.abs.i8(i8 %a36, i1 false)
+  %r37 = call i8 @llvm.abs.i8(i8 %a37, i1 false)
+  %r38 = call i8 @llvm.abs.i8(i8 %a38, i1 false)
+  %r39 = call i8 @llvm.abs.i8(i8 %a39, i1 false)
+  %r40 = call i8 @llvm.abs.i8(i8 %a40, i1 false)
+  %r41 = call i8 @llvm.abs.i8(i8 %a41, i1 false)
+  %r42 = call i8 @llvm.abs.i8(i8 %a42, i1 false)
+  %r43 = call i8 @llvm.abs.i8(i8 %a43, i1 false)
+  %r44 = call i8 @llvm.abs.i8(i8 %a44, i1 false)
+  %r45 = call i8 @llvm.abs.i8(i8 %a45, i1 false)
+  %r46 = call i8 @llvm.abs.i8(i8 %a46, i1 false)
+  %r47 = call i8 @llvm.abs.i8(i8 %a47, i1 false)
+  %r48 = call i8 @llvm.abs.i8(i8 %a48, i1 false)
+  %r49 = call i8 @llvm.abs.i8(i8 %a49, i1 false)
+  %r50 = call i8 @llvm.abs.i8(i8 %a50, i1 false)
+  %r51 = call i8 @llvm.abs.i8(i8 %a51, i1 false)
+  %r52 = call i8 @llvm.abs.i8(i8 %a52, i1 false)
+  %r53 = call i8 @llvm.abs.i8(i8 %a53, i1 false)
+  %r54 = call i8 @llvm.abs.i8(i8 %a54, i1 false)
+  %r55 = call i8 @llvm.abs.i8(i8 %a55, i1 false)
+  %r56 = call i8 @llvm.abs.i8(i8 %a56, i1 false)
+  %r57 = call i8 @llvm.abs.i8(i8 %a57, i1 false)
+  %r58 = call i8 @llvm.abs.i8(i8 %a58, i1 false)
+  %r59 = call i8 @llvm.abs.i8(i8 %a59, i1 false)
+  %r60 = call i8 @llvm.abs.i8(i8 %a60, i1 false)
+  %r61 = call i8 @llvm.abs.i8(i8 %a61, i1 false)
+  %r62 = call i8 @llvm.abs.i8(i8 %a62, i1 false)
+  %r63 = call i8 @llvm.abs.i8(i8 %a63, i1 false)
+  store i8 %r0 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 0 ), align 1
+  store i8 %r1 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 1 ), align 1
+  store i8 %r2 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 2 ), align 1
+  store i8 %r3 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 3 ), align 1
+  store i8 %r4 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 4 ), align 1
+  store i8 %r5 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 5 ), align 1
+  store i8 %r6 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 6 ), align 1
+  store i8 %r7 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 7 ), align 1
+  store i8 %r8 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 8 ), align 1
+  store i8 %r9 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 9 ), align 1
+  store i8 %r10, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 10), align 1
+  store i8 %r11, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 11), align 1
+  store i8 %r12, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 12), align 1
+  store i8 %r13, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 13), align 1
+  store i8 %r14, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 14), align 1
+  store i8 %r15, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 15), align 1
+  store i8 %r16, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16), align 1
+  store i8 %r17, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 17), align 1
+  store i8 %r18, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 18), align 1
+  store i8 %r19, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 19), align 1
+  store i8 %r20, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 20), align 1
+  store i8 %r21, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 21), align 1
+  store i8 %r22, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 22), align 1
+  store i8 %r23, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 23), align 1
+  store i8 %r24, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 24), align 1
+  store i8 %r25, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 25), align 1
+  store i8 %r26, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 26), align 1
+  store i8 %r27, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 27), align 1
+  store i8 %r28, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 28), align 1
+  store i8 %r29, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 29), align 1
+  store i8 %r30, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 30), align 1
+  store i8 %r31, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 31), align 1
+  store i8 %r32, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32), align 1
+  store i8 %r33, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 33), align 1
+  store i8 %r34, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 34), align 1
+  store i8 %r35, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 35), align 1
+  store i8 %r36, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 36), align 1
+  store i8 %r37, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 37), align 1
+  store i8 %r38, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 38), align 1
+  store i8 %r39, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 39), align 1
+  store i8 %r40, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 40), align 1
+  store i8 %r41, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 41), align 1
+  store i8 %r42, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 42), align 1
+  store i8 %r43, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 43), align 1
+  store i8 %r44, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 44), align 1
+  store i8 %r45, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 45), align 1
+  store i8 %r46, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 46), align 1
+  store i8 %r47, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 47), align 1
+  store i8 %r48, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48), align 1
+  store i8 %r49, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 49), align 1
+  store i8 %r50, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 50), align 1
+  store i8 %r51, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 51), align 1
+  store i8 %r52, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 52), align 1
+  store i8 %r53, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 53), align 1
+  store i8 %r54, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 54), align 1
+  store i8 %r55, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 55), align 1
+  store i8 %r56, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 56), align 1
+  store i8 %r57, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 57), align 1
+  store i8 %r58, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 58), align 1
+  store i8 %r59, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 59), align 1
+  store i8 %r60, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 60), align 1
+  store i8 %r61, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 61), align 1
+  store i8 %r62, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 62), align 1
+  store i8 %r63, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 63), align 1
+  ret void
+}
diff --git a/test/Transforms/SLPVectorizer/X86/arith-smax.ll b/test/Transforms/SLPVectorizer/X86/arith-smax.ll
new file mode 100644
index 00000000000..aad8fba28d4
--- /dev/null
+++ b/test/Transforms/SLPVectorizer/X86/arith-smax.ll
@@ -0,0 +1,762 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -mtriple=x86_64-unknown -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,SSE
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=slm -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,SLM
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=corei7-avx -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=knl -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512F
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=-prefer-256-bit -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512BW
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=+prefer-256-bit -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX256BW
+
+@a64 = common global [8 x i64] zeroinitializer, align 64
+@b64 = common global [8 x i64] zeroinitializer, align 64
+@c64 = common global [8 x i64] zeroinitializer, align 64
+@a32 = common global [16 x i32] zeroinitializer, align 64
+@b32 = common global [16 x i32] zeroinitializer, align 64
+@c32 = common global [16 x i32] zeroinitializer, align 64
+@a16 = common global [32 x i16] zeroinitializer, align 64
+@b16 = common global [32 x i16] zeroinitializer, align 64
+@c16 = common global [32 x i16] zeroinitializer, align 64
+@a8  = common global [64 x i8] zeroinitializer, align 64
+@b8  = common global [64 x i8] zeroinitializer, align 64
+@c8  = common global [64 x i8] zeroinitializer, align 64
+
+declare i64 @llvm.smax.i64(i64, i64)
+declare i32 @llvm.smax.i32(i32, i32)
+declare i16 @llvm.smax.i16(i16, i16)
+declare i8  @llvm.smax.i8 (i8 , i8 )
+
+define void @smax_v8i64() {
+; SSE-LABEL: @smax_v8i64(
+; SSE-NEXT:    [[A0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8
+; SSE-NEXT:    [[A1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8
+; SSE-NEXT:    [[A2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8
+; SSE-NEXT:    [[A3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8
+; SSE-NEXT:    [[A4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8
+; SSE-NEXT:    [[A5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8
+; SSE-NEXT:    [[A6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8
+; SSE-NEXT:    [[A7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8
+; SSE-NEXT:    [[B0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8
+; SSE-NEXT:    [[B1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8
+; SSE-NEXT:    [[B2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8
+; SSE-NEXT:    [[B3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8
+; SSE-NEXT:    [[B4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8
+; SSE-NEXT:    [[B5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8
+; SSE-NEXT:    [[B6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8
+; SSE-NEXT:    [[B7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8
+; SSE-NEXT:    [[R0:%.*]] = call i64 @llvm.smax.i64(i64 [[A0]], i64 [[B0]])
+; SSE-NEXT:    [[R1:%.*]] = call i64 @llvm.smax.i64(i64 [[A1]], i64 [[B1]])
+; SSE-NEXT:    [[R2:%.*]] = call i64 @llvm.smax.i64(i64 [[A2]], i64 [[B2]])
+; SSE-NEXT:    [[R3:%.*]] = call i64 @llvm.smax.i64(i64 [[A3]], i64 [[B3]])
+; SSE-NEXT:    [[R4:%.*]] = call i64 @llvm.smax.i64(i64 [[A4]], i64 [[B4]])
+; SSE-NEXT:    [[R5:%.*]] = call i64 @llvm.smax.i64(i64 [[A5]], i64 [[B5]])
+; SSE-NEXT:    [[R6:%.*]] = call i64 @llvm.smax.i64(i64 [[A6]], i64 [[B6]])
+; SSE-NEXT:    [[R7:%.*]] = call i64 @llvm.smax.i64(i64 [[A7]], i64 [[B7]])
+; SSE-NEXT:    store i64 [[R0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8
+; SSE-NEXT:    store i64 [[R1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8
+; SSE-NEXT:    store i64 [[R2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8
+; SSE-NEXT:    store i64 [[R3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8
+; SSE-NEXT:    store i64 [[R4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8
+; SSE-NEXT:    store i64 [[R5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8
+; SSE-NEXT:    store i64 [[R6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8
+; SSE-NEXT:    store i64 [[R7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @smax_v8i64(
+; SLM-NEXT:    [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @a64 to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP2:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP3:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP4:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP5:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @b64 to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP6:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP7:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP8:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP9:%.*]] = call <2 x i64> @llvm.smax.v2i64(<2 x i64> [[TMP1]], <2 x i64> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <2 x i64> @llvm.smax.v2i64(<2 x i64> [[TMP2]], <2 x i64> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <2 x i64> @llvm.smax.v2i64(<2 x i64> [[TMP3]], <2 x i64> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <2 x i64> @llvm.smax.v2i64(<2 x i64> [[TMP4]], <2 x i64> [[TMP8]])
+; SLM-NEXT:    store <2 x i64> [[TMP9]], <2 x i64>* bitcast ([8 x i64]* @c64 to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP10]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP11]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP12]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @smax_v8i64(
+; AVX-NEXT:    [[TMP1:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @a64 to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP2:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP3:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @b64 to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP4:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP5:%.*]] = call <4 x i64> @llvm.smax.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <4 x i64> @llvm.smax.v4i64(<4 x i64> [[TMP2]], <4 x i64> [[TMP4]])
+; AVX-NEXT:    store <4 x i64> [[TMP5]], <4 x i64>* bitcast ([8 x i64]* @c64 to <4 x i64>*), align 8
+; AVX-NEXT:    store <4 x i64> [[TMP6]], <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @smax_v8i64(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <8 x i64>, <8 x i64>* bitcast ([8 x i64]* @a64 to <8 x i64>*), align 8
+; AVX512-NEXT:    [[TMP2:%.*]] = load <8 x i64>, <8 x i64>* bitcast ([8 x i64]* @b64 to <8 x i64>*), align 8
+; AVX512-NEXT:    [[TMP3:%.*]] = call <8 x i64> @llvm.smax.v8i64(<8 x i64> [[TMP1]], <8 x i64> [[TMP2]])
+; AVX512-NEXT:    store <8 x i64> [[TMP3]], <8 x i64>* bitcast ([8 x i64]* @c64 to <8 x i64>*), align 8
+; AVX512-NEXT:    ret void
+;
+  %a0 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8
+  %a1 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8
+  %a2 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8
+  %a3 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8
+  %a4 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8
+  %a5 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8
+  %a6 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8
+  %a7 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8
+  %b0 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8
+  %b1 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8
+  %b2 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8
+  %b3 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8
+  %b4 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8
+  %b5 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8
+  %b6 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8
+  %b7 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8
+  %r0 = call i64 @llvm.smax.i64(i64 %a0, i64 %b0)
+  %r1 = call i64 @llvm.smax.i64(i64 %a1, i64 %b1)
+  %r2 = call i64 @llvm.smax.i64(i64 %a2, i64 %b2)
+  %r3 = call i64 @llvm.smax.i64(i64 %a3, i64 %b3)
+  %r4 = call i64 @llvm.smax.i64(i64 %a4, i64 %b4)
+  %r5 = call i64 @llvm.smax.i64(i64 %a5, i64 %b5)
+  %r6 = call i64 @llvm.smax.i64(i64 %a6, i64 %b6)
+  %r7 = call i64 @llvm.smax.i64(i64 %a7, i64 %b7)
+  store i64 %r0, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8
+  store i64 %r1, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8
+  store i64 %r2, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8
+  store i64 %r3, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8
+  store i64 %r4, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8
+  store i64 %r5, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8
+  store i64 %r6, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8
+  store i64 %r7, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8
+  ret void
+}
+
+define void @smax_v16i32() {
+; SSE-LABEL: @smax_v16i32(
+; SSE-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP5:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @b32 to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP6:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP7:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP8:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP9:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP2]], <4 x i32> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP3]], <4 x i32> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP8]])
+; SSE-NEXT:    store <4 x i32> [[TMP9]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP10]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP11]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP12]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @smax_v16i32(
+; SLM-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP5:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @b32 to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP6:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP7:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP8:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP9:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP2]], <4 x i32> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP3]], <4 x i32> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP8]])
+; SLM-NEXT:    store <4 x i32> [[TMP9]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP10]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP11]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP12]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @smax_v16i32(
+; AVX-NEXT:    [[TMP1:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([16 x i32]* @a32 to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP3:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([16 x i32]* @b32 to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP4:%.*]] = load <8 x i32>, <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP5:%.*]] = call <8 x i32> @llvm.smax.v8i32(<8 x i32> [[TMP1]], <8 x i32> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <8 x i32> @llvm.smax.v8i32(<8 x i32> [[TMP2]], <8 x i32> [[TMP4]])
+; AVX-NEXT:    store <8 x i32> [[TMP5]], <8 x i32>* bitcast ([16 x i32]* @c32 to <8 x i32>*), align 4
+; AVX-NEXT:    store <8 x i32> [[TMP6]], <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @smax_v16i32(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <16 x i32>, <16 x i32>* bitcast ([16 x i32]* @a32 to <16 x i32>*), align 4
+; AVX512-NEXT:    [[TMP2:%.*]] = load <16 x i32>, <16 x i32>* bitcast ([16 x i32]* @b32 to <16 x i32>*), align 4
+; AVX512-NEXT:    [[TMP3:%.*]] = call <16 x i32> @llvm.smax.v16i32(<16 x i32> [[TMP1]], <16 x i32> [[TMP2]])
+; AVX512-NEXT:    store <16 x i32> [[TMP3]], <16 x i32>* bitcast ([16 x i32]* @c32 to <16 x i32>*), align 4
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 0 ), align 4
+  %a1  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 1 ), align 4
+  %a2  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 2 ), align 4
+  %a3  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 3 ), align 4
+  %a4  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4 ), align 4
+  %a5  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 5 ), align 4
+  %a6  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 6 ), align 4
+  %a7  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 7 ), align 4
+  %a8  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8 ), align 4
+  %a9  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 9 ), align 4
+  %a10 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 10), align 4
+  %a11 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 11), align 4
+  %a12 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12), align 4
+  %a13 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 13), align 4
+  %a14 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 14), align 4
+  %a15 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 15), align 4
+  %b0  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 0 ), align 4
+  %b1  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 1 ), align 4
+  %b2  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 2 ), align 4
+  %b3  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 3 ), align 4
+  %b4  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4 ), align 4
+  %b5  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 5 ), align 4
+  %b6  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 6 ), align 4
+  %b7  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 7 ), align 4
+  %b8  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8 ), align 4
+  %b9  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 9 ), align 4
+  %b10 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 10), align 4
+  %b11 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 11), align 4
+  %b12 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12), align 4
+  %b13 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 13), align 4
+  %b14 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 14), align 4
+  %b15 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 15), align 4
+  %r0  = call i32 @llvm.smax.i32(i32 %a0 , i32 %b0 )
+  %r1  = call i32 @llvm.smax.i32(i32 %a1 , i32 %b1 )
+  %r2  = call i32 @llvm.smax.i32(i32 %a2 , i32 %b2 )
+  %r3  = call i32 @llvm.smax.i32(i32 %a3 , i32 %b3 )
+  %r4  = call i32 @llvm.smax.i32(i32 %a4 , i32 %b4 )
+  %r5  = call i32 @llvm.smax.i32(i32 %a5 , i32 %b5 )
+  %r6  = call i32 @llvm.smax.i32(i32 %a6 , i32 %b6 )
+  %r7  = call i32 @llvm.smax.i32(i32 %a7 , i32 %b7 )
+  %r8  = call i32 @llvm.smax.i32(i32 %a8 , i32 %b8 )
+  %r9  = call i32 @llvm.smax.i32(i32 %a9 , i32 %b9 )
+  %r10 = call i32 @llvm.smax.i32(i32 %a10, i32 %b10)
+  %r11 = call i32 @llvm.smax.i32(i32 %a11, i32 %b11)
+  %r12 = call i32 @llvm.smax.i32(i32 %a12, i32 %b12)
+  %r13 = call i32 @llvm.smax.i32(i32 %a13, i32 %b13)
+  %r14 = call i32 @llvm.smax.i32(i32 %a14, i32 %b14)
+  %r15 = call i32 @llvm.smax.i32(i32 %a15, i32 %b15)
+  store i32 %r0 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 0 ), align 4
+  store i32 %r1 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 1 ), align 4
+  store i32 %r2 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 2 ), align 4
+  store i32 %r3 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 3 ), align 4
+  store i32 %r4 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4 ), align 4
+  store i32 %r5 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 5 ), align 4
+  store i32 %r6 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 6 ), align 4
+  store i32 %r7 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 7 ), align 4
+  store i32 %r8 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8 ), align 4
+  store i32 %r9 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 9 ), align 4
+  store i32 %r10, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 10), align 4
+  store i32 %r11, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 11), align 4
+  store i32 %r12, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12), align 4
+  store i32 %r13, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 13), align 4
+  store i32 %r14, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 14), align 4
+  store i32 %r15, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 15), align 4
+  ret void
+}
+
+define void @smax_v32i16() {
+; SSE-LABEL: @smax_v32i16(
+; SSE-NEXT:    [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @a16 to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP3:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP5:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @b16 to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP6:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP7:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP8:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP9:%.*]] = call <8 x i16> @llvm.smax.v8i16(<8 x i16> [[TMP1]], <8 x i16> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <8 x i16> @llvm.smax.v8i16(<8 x i16> [[TMP2]], <8 x i16> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <8 x i16> @llvm.smax.v8i16(<8 x i16> [[TMP3]], <8 x i16> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <8 x i16> @llvm.smax.v8i16(<8 x i16> [[TMP4]], <8 x i16> [[TMP8]])
+; SSE-NEXT:    store <8 x i16> [[TMP9]], <8 x i16>* bitcast ([32 x i16]* @c16 to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP10]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP11]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP12]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @smax_v32i16(
+; SLM-NEXT:    [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @a16 to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP3:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP5:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @b16 to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP6:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP7:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP8:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP9:%.*]] = call <8 x i16> @llvm.smax.v8i16(<8 x i16> [[TMP1]], <8 x i16> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <8 x i16> @llvm.smax.v8i16(<8 x i16> [[TMP2]], <8 x i16> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <8 x i16> @llvm.smax.v8i16(<8 x i16> [[TMP3]], <8 x i16> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <8 x i16> @llvm.smax.v8i16(<8 x i16> [[TMP4]], <8 x i16> [[TMP8]])
+; SLM-NEXT:    store <8 x i16> [[TMP9]], <8 x i16>* bitcast ([32 x i16]* @c16 to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP10]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP11]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP12]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @smax_v32i16(
+; AVX-NEXT:    [[TMP1:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([32 x i16]* @a16 to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP2:%.*]] = load <16 x i16>, <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP3:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([32 x i16]* @b16 to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP4:%.*]] = load <16 x i16>, <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP5:%.*]] = call <16 x i16> @llvm.smax.v16i16(<16 x i16> [[TMP1]], <16 x i16> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <16 x i16> @llvm.smax.v16i16(<16 x i16> [[TMP2]], <16 x i16> [[TMP4]])
+; AVX-NEXT:    store <16 x i16> [[TMP5]], <16 x i16>* bitcast ([32 x i16]* @c16 to <16 x i16>*), align 2
+; AVX-NEXT:    store <16 x i16> [[TMP6]], <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @smax_v32i16(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <32 x i16>, <32 x i16>* bitcast ([32 x i16]* @a16 to <32 x i16>*), align 2
+; AVX512-NEXT:    [[TMP2:%.*]] = load <32 x i16>, <32 x i16>* bitcast ([32 x i16]* @b16 to <32 x i16>*), align 2
+; AVX512-NEXT:    [[TMP3:%.*]] = call <32 x i16> @llvm.smax.v32i16(<32 x i16> [[TMP1]], <32 x i16> [[TMP2]])
+; AVX512-NEXT:    store <32 x i16> [[TMP3]], <32 x i16>* bitcast ([32 x i16]* @c16 to <32 x i16>*), align 2
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 0 ), align 2
+  %a1  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 1 ), align 2
+  %a2  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 2 ), align 2
+  %a3  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 3 ), align 2
+  %a4  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 4 ), align 2
+  %a5  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 5 ), align 2
+  %a6  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 6 ), align 2
+  %a7  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 7 ), align 2
+  %a8  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8 ), align 2
+  %a9  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 9 ), align 2
+  %a10 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 10), align 2
+  %a11 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 11), align 2
+  %a12 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 12), align 2
+  %a13 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 13), align 2
+  %a14 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 14), align 2
+  %a15 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 15), align 2
+  %a16 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16), align 2
+  %a17 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 17), align 2
+  %a18 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 18), align 2
+  %a19 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 19), align 2
+  %a20 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 20), align 2
+  %a21 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 21), align 2
+  %a22 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 22), align 2
+  %a23 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 23), align 2
+  %a24 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24), align 2
+  %a25 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 25), align 2
+  %a26 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 26), align 2
+  %a27 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 27), align 2
+  %a28 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 28), align 2
+  %a29 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 29), align 2
+  %a30 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 30), align 2
+  %a31 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 31), align 2
+  %b0  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 0 ), align 2
+  %b1  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 1 ), align 2
+  %b2  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 2 ), align 2
+  %b3  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 3 ), align 2
+  %b4  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 4 ), align 2
+  %b5  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 5 ), align 2
+  %b6  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 6 ), align 2
+  %b7  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 7 ), align 2
+  %b8  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8 ), align 2
+  %b9  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 9 ), align 2
+  %b10 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 10), align 2
+  %b11 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 11), align 2
+  %b12 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 12), align 2
+  %b13 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 13), align 2
+  %b14 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 14), align 2
+  %b15 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 15), align 2
+  %b16 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16), align 2
+  %b17 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 17), align 2
+  %b18 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 18), align 2
+  %b19 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 19), align 2
+  %b20 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 20), align 2
+  %b21 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 21), align 2
+  %b22 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 22), align 2
+  %b23 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 23), align 2
+  %b24 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24), align 2
+  %b25 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 25), align 2
+  %b26 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 26), align 2
+  %b27 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 27), align 2
+  %b28 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 28), align 2
+  %b29 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 29), align 2
+  %b30 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 30), align 2
+  %b31 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 31), align 2
+  %r0  = call i16 @llvm.smax.i16(i16 %a0 , i16 %b0 )
+  %r1  = call i16 @llvm.smax.i16(i16 %a1 , i16 %b1 )
+  %r2  = call i16 @llvm.smax.i16(i16 %a2 , i16 %b2 )
+  %r3  = call i16 @llvm.smax.i16(i16 %a3 , i16 %b3 )
+  %r4  = call i16 @llvm.smax.i16(i16 %a4 , i16 %b4 )
+  %r5  = call i16 @llvm.smax.i16(i16 %a5 , i16 %b5 )
+  %r6  = call i16 @llvm.smax.i16(i16 %a6 , i16 %b6 )
+  %r7  = call i16 @llvm.smax.i16(i16 %a7 , i16 %b7 )
+  %r8  = call i16 @llvm.smax.i16(i16 %a8 , i16 %b8 )
+  %r9  = call i16 @llvm.smax.i16(i16 %a9 , i16 %b9 )
+  %r10 = call i16 @llvm.smax.i16(i16 %a10, i16 %b10)
+  %r11 = call i16 @llvm.smax.i16(i16 %a11, i16 %b11)
+  %r12 = call i16 @llvm.smax.i16(i16 %a12, i16 %b12)
+  %r13 = call i16 @llvm.smax.i16(i16 %a13, i16 %b13)
+  %r14 = call i16 @llvm.smax.i16(i16 %a14, i16 %b14)
+  %r15 = call i16 @llvm.smax.i16(i16 %a15, i16 %b15)
+  %r16 = call i16 @llvm.smax.i16(i16 %a16, i16 %b16)
+  %r17 = call i16 @llvm.smax.i16(i16 %a17, i16 %b17)
+  %r18 = call i16 @llvm.smax.i16(i16 %a18, i16 %b18)
+  %r19 = call i16 @llvm.smax.i16(i16 %a19, i16 %b19)
+  %r20 = call i16 @llvm.smax.i16(i16 %a20, i16 %b20)
+  %r21 = call i16 @llvm.smax.i16(i16 %a21, i16 %b21)
+  %r22 = call i16 @llvm.smax.i16(i16 %a22, i16 %b22)
+  %r23 = call i16 @llvm.smax.i16(i16 %a23, i16 %b23)
+  %r24 = call i16 @llvm.smax.i16(i16 %a24, i16 %b24)
+  %r25 = call i16 @llvm.smax.i16(i16 %a25, i16 %b25)
+  %r26 = call i16 @llvm.smax.i16(i16 %a26, i16 %b26)
+  %r27 = call i16 @llvm.smax.i16(i16 %a27, i16 %b27)
+  %r28 = call i16 @llvm.smax.i16(i16 %a28, i16 %b28)
+  %r29 = call i16 @llvm.smax.i16(i16 %a29, i16 %b29)
+  %r30 = call i16 @llvm.smax.i16(i16 %a30, i16 %b30)
+  %r31 = call i16 @llvm.smax.i16(i16 %a31, i16 %b31)
+  store i16 %r0 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 0 ), align 2
+  store i16 %r1 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 1 ), align 2
+  store i16 %r2 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 2 ), align 2
+  store i16 %r3 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 3 ), align 2
+  store i16 %r4 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 4 ), align 2
+  store i16 %r5 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 5 ), align 2
+  store i16 %r6 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 6 ), align 2
+  store i16 %r7 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 7 ), align 2
+  store i16 %r8 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8 ), align 2
+  store i16 %r9 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 9 ), align 2
+  store i16 %r10, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 10), align 2
+  store i16 %r11, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 11), align 2
+  store i16 %r12, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 12), align 2
+  store i16 %r13, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 13), align 2
+  store i16 %r14, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 14), align 2
+  store i16 %r15, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 15), align 2
+  store i16 %r16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16), align 2
+  store i16 %r17, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 17), align 2
+  store i16 %r18, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 18), align 2
+  store i16 %r19, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 19), align 2
+  store i16 %r20, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 20), align 2
+  store i16 %r21, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 21), align 2
+  store i16 %r22, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 22), align 2
+  store i16 %r23, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 23), align 2
+  store i16 %r24, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24), align 2
+  store i16 %r25, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 25), align 2
+  store i16 %r26, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 26), align 2
+  store i16 %r27, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 27), align 2
+  store i16 %r28, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 28), align 2
+  store i16 %r29, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 29), align 2
+  store i16 %r30, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 30), align 2
+  store i16 %r31, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 31), align 2
+  ret void
+}
+
+define void @smax_v64i8() {
+; SSE-LABEL: @smax_v64i8(
+; SSE-NEXT:    [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @a8 to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP5:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @b8 to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP6:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP7:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP8:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP9:%.*]] = call <16 x i8> @llvm.smax.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <16 x i8> @llvm.smax.v16i8(<16 x i8> [[TMP2]], <16 x i8> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <16 x i8> @llvm.smax.v16i8(<16 x i8> [[TMP3]], <16 x i8> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <16 x i8> @llvm.smax.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP8]])
+; SSE-NEXT:    store <16 x i8> [[TMP9]], <16 x i8>* bitcast ([64 x i8]* @c8 to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP10]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP11]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP12]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @smax_v64i8(
+; SLM-NEXT:    [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @a8 to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP5:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @b8 to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP6:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP7:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP8:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP9:%.*]] = call <16 x i8> @llvm.smax.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <16 x i8> @llvm.smax.v16i8(<16 x i8> [[TMP2]], <16 x i8> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <16 x i8> @llvm.smax.v16i8(<16 x i8> [[TMP3]], <16 x i8> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <16 x i8> @llvm.smax.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP8]])
+; SLM-NEXT:    store <16 x i8> [[TMP9]], <16 x i8>* bitcast ([64 x i8]* @c8 to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP10]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP11]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP12]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @smax_v64i8(
+; AVX-NEXT:    [[TMP1:%.*]] = load <32 x i8>, <32 x i8>* bitcast ([64 x i8]* @a8 to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP2:%.*]] = load <32 x i8>, <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP3:%.*]] = load <32 x i8>, <32 x i8>* bitcast ([64 x i8]* @b8 to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP4:%.*]] = load <32 x i8>, <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP5:%.*]] = call <32 x i8> @llvm.smax.v32i8(<32 x i8> [[TMP1]], <32 x i8> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <32 x i8> @llvm.smax.v32i8(<32 x i8> [[TMP2]], <32 x i8> [[TMP4]])
+; AVX-NEXT:    store <32 x i8> [[TMP5]], <32 x i8>* bitcast ([64 x i8]* @c8 to <32 x i8>*), align 1
+; AVX-NEXT:    store <32 x i8> [[TMP6]], <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @smax_v64i8(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <64 x i8>, <64 x i8>* bitcast ([64 x i8]* @a8 to <64 x i8>*), align 1
+; AVX512-NEXT:    [[TMP2:%.*]] = load <64 x i8>, <64 x i8>* bitcast ([64 x i8]* @b8 to <64 x i8>*), align 1
+; AVX512-NEXT:    [[TMP3:%.*]] = call <64 x i8> @llvm.smax.v64i8(<64 x i8> [[TMP1]], <64 x i8> [[TMP2]])
+; AVX512-NEXT:    store <64 x i8> [[TMP3]], <64 x i8>* bitcast ([64 x i8]* @c8 to <64 x i8>*), align 1
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 0 ), align 1
+  %a1  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 1 ), align 1
+  %a2  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 2 ), align 1
+  %a3  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 3 ), align 1
+  %a4  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 4 ), align 1
+  %a5  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 5 ), align 1
+  %a6  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 6 ), align 1
+  %a7  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 7 ), align 1
+  %a8  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 8 ), align 1
+  %a9  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 9 ), align 1
+  %a10 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 10), align 1
+  %a11 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 11), align 1
+  %a12 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 12), align 1
+  %a13 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 13), align 1
+  %a14 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 14), align 1
+  %a15 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 15), align 1
+  %a16 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16), align 1
+  %a17 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 17), align 1
+  %a18 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 18), align 1
+  %a19 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 19), align 1
+  %a20 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 20), align 1
+  %a21 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 21), align 1
+  %a22 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 22), align 1
+  %a23 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 23), align 1
+  %a24 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 24), align 1
+  %a25 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 25), align 1
+  %a26 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 26), align 1
+  %a27 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 27), align 1
+  %a28 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 28), align 1
+  %a29 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 29), align 1
+  %a30 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 30), align 1
+  %a31 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 31), align 1
+  %a32 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32), align 1
+  %a33 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 33), align 1
+  %a34 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 34), align 1
+  %a35 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 35), align 1
+  %a36 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 36), align 1
+  %a37 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 37), align 1
+  %a38 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 38), align 1
+  %a39 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 39), align 1
+  %a40 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 40), align 1
+  %a41 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 41), align 1
+  %a42 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 42), align 1
+  %a43 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 43), align 1
+  %a44 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 44), align 1
+  %a45 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 45), align 1
+  %a46 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 46), align 1
+  %a47 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 47), align 1
+  %a48 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48), align 1
+  %a49 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 49), align 1
+  %a50 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 50), align 1
+  %a51 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 51), align 1
+  %a52 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 52), align 1
+  %a53 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 53), align 1
+  %a54 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 54), align 1
+  %a55 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 55), align 1
+  %a56 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 56), align 1
+  %a57 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 57), align 1
+  %a58 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 58), align 1
+  %a59 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 59), align 1
+  %a60 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 60), align 1
+  %a61 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 61), align 1
+  %a62 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 62), align 1
+  %a63 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 63), align 1
+  %b0  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 0 ), align 1
+  %b1  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 1 ), align 1
+  %b2  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 2 ), align 1
+  %b3  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 3 ), align 1
+  %b4  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 4 ), align 1
+  %b5  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 5 ), align 1
+  %b6  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 6 ), align 1
+  %b7  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 7 ), align 1
+  %b8  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 8 ), align 1
+  %b9  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 9 ), align 1
+  %b10 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 10), align 1
+  %b11 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 11), align 1
+  %b12 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 12), align 1
+  %b13 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 13), align 1
+  %b14 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 14), align 1
+  %b15 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 15), align 1
+  %b16 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16), align 1
+  %b17 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 17), align 1
+  %b18 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 18), align 1
+  %b19 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 19), align 1
+  %b20 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 20), align 1
+  %b21 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 21), align 1
+  %b22 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 22), align 1
+  %b23 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 23), align 1
+  %b24 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 24), align 1
+  %b25 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 25), align 1
+  %b26 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 26), align 1
+  %b27 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 27), align 1
+  %b28 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 28), align 1
+  %b29 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 29), align 1
+  %b30 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 30), align 1
+  %b31 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 31), align 1
+  %b32 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32), align 1
+  %b33 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 33), align 1
+  %b34 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 34), align 1
+  %b35 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 35), align 1
+  %b36 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 36), align 1
+  %b37 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 37), align 1
+  %b38 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 38), align 1
+  %b39 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 39), align 1
+  %b40 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 40), align 1
+  %b41 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 41), align 1
+  %b42 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 42), align 1
+  %b43 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 43), align 1
+  %b44 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 44), align 1
+  %b45 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 45), align 1
+  %b46 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 46), align 1
+  %b47 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 47), align 1
+  %b48 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48), align 1
+  %b49 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 49), align 1
+  %b50 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 50), align 1
+  %b51 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 51), align 1
+  %b52 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 52), align 1
+  %b53 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 53), align 1
+  %b54 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 54), align 1
+  %b55 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 55), align 1
+  %b56 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 56), align 1
+  %b57 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 57), align 1
+  %b58 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 58), align 1
+  %b59 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 59), align 1
+  %b60 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 60), align 1
+  %b61 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 61), align 1
+  %b62 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 62), align 1
+  %b63 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 63), align 1
+  %r0  = call i8 @llvm.smax.i8(i8 %a0 , i8 %b0 )
+  %r1  = call i8 @llvm.smax.i8(i8 %a1 , i8 %b1 )
+  %r2  = call i8 @llvm.smax.i8(i8 %a2 , i8 %b2 )
+  %r3  = call i8 @llvm.smax.i8(i8 %a3 , i8 %b3 )
+  %r4  = call i8 @llvm.smax.i8(i8 %a4 , i8 %b4 )
+  %r5  = call i8 @llvm.smax.i8(i8 %a5 , i8 %b5 )
+  %r6  = call i8 @llvm.smax.i8(i8 %a6 , i8 %b6 )
+  %r7  = call i8 @llvm.smax.i8(i8 %a7 , i8 %b7 )
+  %r8  = call i8 @llvm.smax.i8(i8 %a8 , i8 %b8 )
+  %r9  = call i8 @llvm.smax.i8(i8 %a9 , i8 %b9 )
+  %r10 = call i8 @llvm.smax.i8(i8 %a10, i8 %b10)
+  %r11 = call i8 @llvm.smax.i8(i8 %a11, i8 %b11)
+  %r12 = call i8 @llvm.smax.i8(i8 %a12, i8 %b12)
+  %r13 = call i8 @llvm.smax.i8(i8 %a13, i8 %b13)
+  %r14 = call i8 @llvm.smax.i8(i8 %a14, i8 %b14)
+  %r15 = call i8 @llvm.smax.i8(i8 %a15, i8 %b15)
+  %r16 = call i8 @llvm.smax.i8(i8 %a16, i8 %b16)
+  %r17 = call i8 @llvm.smax.i8(i8 %a17, i8 %b17)
+  %r18 = call i8 @llvm.smax.i8(i8 %a18, i8 %b18)
+  %r19 = call i8 @llvm.smax.i8(i8 %a19, i8 %b19)
+  %r20 = call i8 @llvm.smax.i8(i8 %a20, i8 %b20)
+  %r21 = call i8 @llvm.smax.i8(i8 %a21, i8 %b21)
+  %r22 = call i8 @llvm.smax.i8(i8 %a22, i8 %b22)
+  %r23 = call i8 @llvm.smax.i8(i8 %a23, i8 %b23)
+  %r24 = call i8 @llvm.smax.i8(i8 %a24, i8 %b24)
+  %r25 = call i8 @llvm.smax.i8(i8 %a25, i8 %b25)
+  %r26 = call i8 @llvm.smax.i8(i8 %a26, i8 %b26)
+  %r27 = call i8 @llvm.smax.i8(i8 %a27, i8 %b27)
+  %r28 = call i8 @llvm.smax.i8(i8 %a28, i8 %b28)
+  %r29 = call i8 @llvm.smax.i8(i8 %a29, i8 %b29)
+  %r30 = call i8 @llvm.smax.i8(i8 %a30, i8 %b30)
+  %r31 = call i8 @llvm.smax.i8(i8 %a31, i8 %b31)
+  %r32 = call i8 @llvm.smax.i8(i8 %a32, i8 %b32)
+  %r33 = call i8 @llvm.smax.i8(i8 %a33, i8 %b33)
+  %r34 = call i8 @llvm.smax.i8(i8 %a34, i8 %b34)
+  %r35 = call i8 @llvm.smax.i8(i8 %a35, i8 %b35)
+  %r36 = call i8 @llvm.smax.i8(i8 %a36, i8 %b36)
+  %r37 = call i8 @llvm.smax.i8(i8 %a37, i8 %b37)
+  %r38 = call i8 @llvm.smax.i8(i8 %a38, i8 %b38)
+  %r39 = call i8 @llvm.smax.i8(i8 %a39, i8 %b39)
+  %r40 = call i8 @llvm.smax.i8(i8 %a40, i8 %b40)
+  %r41 = call i8 @llvm.smax.i8(i8 %a41, i8 %b41)
+  %r42 = call i8 @llvm.smax.i8(i8 %a42, i8 %b42)
+  %r43 = call i8 @llvm.smax.i8(i8 %a43, i8 %b43)
+  %r44 = call i8 @llvm.smax.i8(i8 %a44, i8 %b44)
+  %r45 = call i8 @llvm.smax.i8(i8 %a45, i8 %b45)
+  %r46 = call i8 @llvm.smax.i8(i8 %a46, i8 %b46)
+  %r47 = call i8 @llvm.smax.i8(i8 %a47, i8 %b47)
+  %r48 = call i8 @llvm.smax.i8(i8 %a48, i8 %b48)
+  %r49 = call i8 @llvm.smax.i8(i8 %a49, i8 %b49)
+  %r50 = call i8 @llvm.smax.i8(i8 %a50, i8 %b50)
+  %r51 = call i8 @llvm.smax.i8(i8 %a51, i8 %b51)
+  %r52 = call i8 @llvm.smax.i8(i8 %a52, i8 %b52)
+  %r53 = call i8 @llvm.smax.i8(i8 %a53, i8 %b53)
+  %r54 = call i8 @llvm.smax.i8(i8 %a54, i8 %b54)
+  %r55 = call i8 @llvm.smax.i8(i8 %a55, i8 %b55)
+  %r56 = call i8 @llvm.smax.i8(i8 %a56, i8 %b56)
+  %r57 = call i8 @llvm.smax.i8(i8 %a57, i8 %b57)
+  %r58 = call i8 @llvm.smax.i8(i8 %a58, i8 %b58)
+  %r59 = call i8 @llvm.smax.i8(i8 %a59, i8 %b59)
+  %r60 = call i8 @llvm.smax.i8(i8 %a60, i8 %b60)
+  %r61 = call i8 @llvm.smax.i8(i8 %a61, i8 %b61)
+  %r62 = call i8 @llvm.smax.i8(i8 %a62, i8 %b62)
+  %r63 = call i8 @llvm.smax.i8(i8 %a63, i8 %b63)
+  store i8 %r0 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 0 ), align 1
+  store i8 %r1 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 1 ), align 1
+  store i8 %r2 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 2 ), align 1
+  store i8 %r3 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 3 ), align 1
+  store i8 %r4 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 4 ), align 1
+  store i8 %r5 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 5 ), align 1
+  store i8 %r6 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 6 ), align 1
+  store i8 %r7 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 7 ), align 1
+  store i8 %r8 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 8 ), align 1
+  store i8 %r9 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 9 ), align 1
+  store i8 %r10, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 10), align 1
+  store i8 %r11, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 11), align 1
+  store i8 %r12, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 12), align 1
+  store i8 %r13, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 13), align 1
+  store i8 %r14, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 14), align 1
+  store i8 %r15, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 15), align 1
+  store i8 %r16, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16), align 1
+  store i8 %r17, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 17), align 1
+  store i8 %r18, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 18), align 1
+  store i8 %r19, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 19), align 1
+  store i8 %r20, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 20), align 1
+  store i8 %r21, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 21), align 1
+  store i8 %r22, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 22), align 1
+  store i8 %r23, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 23), align 1
+  store i8 %r24, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 24), align 1
+  store i8 %r25, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 25), align 1
+  store i8 %r26, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 26), align 1
+  store i8 %r27, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 27), align 1
+  store i8 %r28, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 28), align 1
+  store i8 %r29, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 29), align 1
+  store i8 %r30, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 30), align 1
+  store i8 %r31, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 31), align 1
+  store i8 %r32, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32), align 1
+  store i8 %r33, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 33), align 1
+  store i8 %r34, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 34), align 1
+  store i8 %r35, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 35), align 1
+  store i8 %r36, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 36), align 1
+  store i8 %r37, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 37), align 1
+  store i8 %r38, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 38), align 1
+  store i8 %r39, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 39), align 1
+  store i8 %r40, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 40), align 1
+  store i8 %r41, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 41), align 1
+  store i8 %r42, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 42), align 1
+  store i8 %r43, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 43), align 1
+  store i8 %r44, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 44), align 1
+  store i8 %r45, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 45), align 1
+  store i8 %r46, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 46), align 1
+  store i8 %r47, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 47), align 1
+  store i8 %r48, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48), align 1
+  store i8 %r49, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 49), align 1
+  store i8 %r50, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 50), align 1
+  store i8 %r51, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 51), align 1
+  store i8 %r52, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 52), align 1
+  store i8 %r53, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 53), align 1
+  store i8 %r54, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 54), align 1
+  store i8 %r55, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 55), align 1
+  store i8 %r56, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 56), align 1
+  store i8 %r57, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 57), align 1
+  store i8 %r58, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 58), align 1
+  store i8 %r59, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 59), align 1
+  store i8 %r60, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 60), align 1
+  store i8 %r61, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 61), align 1
+  store i8 %r62, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 62), align 1
+  store i8 %r63, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 63), align 1
+  ret void
+}
diff --git a/test/Transforms/SLPVectorizer/X86/arith-smin.ll b/test/Transforms/SLPVectorizer/X86/arith-smin.ll
new file mode 100644
index 00000000000..e0e19d029bd
--- /dev/null
+++ b/test/Transforms/SLPVectorizer/X86/arith-smin.ll
@@ -0,0 +1,762 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -mtriple=x86_64-unknown -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,SSE
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=slm -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,SLM
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=corei7-avx -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=knl -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512F
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=-prefer-256-bit -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512BW
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=+prefer-256-bit -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX256BW
+
+@a64 = common global [8 x i64] zeroinitializer, align 64
+@b64 = common global [8 x i64] zeroinitializer, align 64
+@c64 = common global [8 x i64] zeroinitializer, align 64
+@a32 = common global [16 x i32] zeroinitializer, align 64
+@b32 = common global [16 x i32] zeroinitializer, align 64
+@c32 = common global [16 x i32] zeroinitializer, align 64
+@a16 = common global [32 x i16] zeroinitializer, align 64
+@b16 = common global [32 x i16] zeroinitializer, align 64
+@c16 = common global [32 x i16] zeroinitializer, align 64
+@a8  = common global [64 x i8] zeroinitializer, align 64
+@b8  = common global [64 x i8] zeroinitializer, align 64
+@c8  = common global [64 x i8] zeroinitializer, align 64
+
+declare i64 @llvm.smin.i64(i64, i64)
+declare i32 @llvm.smin.i32(i32, i32)
+declare i16 @llvm.smin.i16(i16, i16)
+declare i8  @llvm.smin.i8 (i8 , i8 )
+
+define void @smin_v8i64() {
+; SSE-LABEL: @smin_v8i64(
+; SSE-NEXT:    [[A0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8
+; SSE-NEXT:    [[A1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8
+; SSE-NEXT:    [[A2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8
+; SSE-NEXT:    [[A3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8
+; SSE-NEXT:    [[A4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8
+; SSE-NEXT:    [[A5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8
+; SSE-NEXT:    [[A6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8
+; SSE-NEXT:    [[A7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8
+; SSE-NEXT:    [[B0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8
+; SSE-NEXT:    [[B1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8
+; SSE-NEXT:    [[B2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8
+; SSE-NEXT:    [[B3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8
+; SSE-NEXT:    [[B4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8
+; SSE-NEXT:    [[B5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8
+; SSE-NEXT:    [[B6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8
+; SSE-NEXT:    [[B7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8
+; SSE-NEXT:    [[R0:%.*]] = call i64 @llvm.smin.i64(i64 [[A0]], i64 [[B0]])
+; SSE-NEXT:    [[R1:%.*]] = call i64 @llvm.smin.i64(i64 [[A1]], i64 [[B1]])
+; SSE-NEXT:    [[R2:%.*]] = call i64 @llvm.smin.i64(i64 [[A2]], i64 [[B2]])
+; SSE-NEXT:    [[R3:%.*]] = call i64 @llvm.smin.i64(i64 [[A3]], i64 [[B3]])
+; SSE-NEXT:    [[R4:%.*]] = call i64 @llvm.smin.i64(i64 [[A4]], i64 [[B4]])
+; SSE-NEXT:    [[R5:%.*]] = call i64 @llvm.smin.i64(i64 [[A5]], i64 [[B5]])
+; SSE-NEXT:    [[R6:%.*]] = call i64 @llvm.smin.i64(i64 [[A6]], i64 [[B6]])
+; SSE-NEXT:    [[R7:%.*]] = call i64 @llvm.smin.i64(i64 [[A7]], i64 [[B7]])
+; SSE-NEXT:    store i64 [[R0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8
+; SSE-NEXT:    store i64 [[R1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8
+; SSE-NEXT:    store i64 [[R2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8
+; SSE-NEXT:    store i64 [[R3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8
+; SSE-NEXT:    store i64 [[R4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8
+; SSE-NEXT:    store i64 [[R5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8
+; SSE-NEXT:    store i64 [[R6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8
+; SSE-NEXT:    store i64 [[R7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @smin_v8i64(
+; SLM-NEXT:    [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @a64 to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP2:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP3:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP4:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP5:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @b64 to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP6:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP7:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP8:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP9:%.*]] = call <2 x i64> @llvm.smin.v2i64(<2 x i64> [[TMP1]], <2 x i64> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <2 x i64> @llvm.smin.v2i64(<2 x i64> [[TMP2]], <2 x i64> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <2 x i64> @llvm.smin.v2i64(<2 x i64> [[TMP3]], <2 x i64> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <2 x i64> @llvm.smin.v2i64(<2 x i64> [[TMP4]], <2 x i64> [[TMP8]])
+; SLM-NEXT:    store <2 x i64> [[TMP9]], <2 x i64>* bitcast ([8 x i64]* @c64 to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP10]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP11]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP12]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @smin_v8i64(
+; AVX-NEXT:    [[TMP1:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @a64 to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP2:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP3:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @b64 to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP4:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP5:%.*]] = call <4 x i64> @llvm.smin.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <4 x i64> @llvm.smin.v4i64(<4 x i64> [[TMP2]], <4 x i64> [[TMP4]])
+; AVX-NEXT:    store <4 x i64> [[TMP5]], <4 x i64>* bitcast ([8 x i64]* @c64 to <4 x i64>*), align 8
+; AVX-NEXT:    store <4 x i64> [[TMP6]], <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @smin_v8i64(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <8 x i64>, <8 x i64>* bitcast ([8 x i64]* @a64 to <8 x i64>*), align 8
+; AVX512-NEXT:    [[TMP2:%.*]] = load <8 x i64>, <8 x i64>* bitcast ([8 x i64]* @b64 to <8 x i64>*), align 8
+; AVX512-NEXT:    [[TMP3:%.*]] = call <8 x i64> @llvm.smin.v8i64(<8 x i64> [[TMP1]], <8 x i64> [[TMP2]])
+; AVX512-NEXT:    store <8 x i64> [[TMP3]], <8 x i64>* bitcast ([8 x i64]* @c64 to <8 x i64>*), align 8
+; AVX512-NEXT:    ret void
+;
+  %a0 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8
+  %a1 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8
+  %a2 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8
+  %a3 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8
+  %a4 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8
+  %a5 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8
+  %a6 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8
+  %a7 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8
+  %b0 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8
+  %b1 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8
+  %b2 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8
+  %b3 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8
+  %b4 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8
+  %b5 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8
+  %b6 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8
+  %b7 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8
+  %r0 = call i64 @llvm.smin.i64(i64 %a0, i64 %b0)
+  %r1 = call i64 @llvm.smin.i64(i64 %a1, i64 %b1)
+  %r2 = call i64 @llvm.smin.i64(i64 %a2, i64 %b2)
+  %r3 = call i64 @llvm.smin.i64(i64 %a3, i64 %b3)
+  %r4 = call i64 @llvm.smin.i64(i64 %a4, i64 %b4)
+  %r5 = call i64 @llvm.smin.i64(i64 %a5, i64 %b5)
+  %r6 = call i64 @llvm.smin.i64(i64 %a6, i64 %b6)
+  %r7 = call i64 @llvm.smin.i64(i64 %a7, i64 %b7)
+  store i64 %r0, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8
+  store i64 %r1, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8
+  store i64 %r2, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8
+  store i64 %r3, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8
+  store i64 %r4, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8
+  store i64 %r5, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8
+  store i64 %r6, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8
+  store i64 %r7, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8
+  ret void
+}
+
+define void @smin_v16i32() {
+; SSE-LABEL: @smin_v16i32(
+; SSE-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP5:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @b32 to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP6:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP7:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP8:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP9:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[TMP2]], <4 x i32> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[TMP3]], <4 x i32> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP8]])
+; SSE-NEXT:    store <4 x i32> [[TMP9]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP10]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP11]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP12]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @smin_v16i32(
+; SLM-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP5:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @b32 to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP6:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP7:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP8:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP9:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[TMP2]], <4 x i32> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[TMP3]], <4 x i32> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP8]])
+; SLM-NEXT:    store <4 x i32> [[TMP9]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP10]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP11]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP12]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @smin_v16i32(
+; AVX-NEXT:    [[TMP1:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([16 x i32]* @a32 to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP3:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([16 x i32]* @b32 to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP4:%.*]] = load <8 x i32>, <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP5:%.*]] = call <8 x i32> @llvm.smin.v8i32(<8 x i32> [[TMP1]], <8 x i32> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <8 x i32> @llvm.smin.v8i32(<8 x i32> [[TMP2]], <8 x i32> [[TMP4]])
+; AVX-NEXT:    store <8 x i32> [[TMP5]], <8 x i32>* bitcast ([16 x i32]* @c32 to <8 x i32>*), align 4
+; AVX-NEXT:    store <8 x i32> [[TMP6]], <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @smin_v16i32(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <16 x i32>, <16 x i32>* bitcast ([16 x i32]* @a32 to <16 x i32>*), align 4
+; AVX512-NEXT:    [[TMP2:%.*]] = load <16 x i32>, <16 x i32>* bitcast ([16 x i32]* @b32 to <16 x i32>*), align 4
+; AVX512-NEXT:    [[TMP3:%.*]] = call <16 x i32> @llvm.smin.v16i32(<16 x i32> [[TMP1]], <16 x i32> [[TMP2]])
+; AVX512-NEXT:    store <16 x i32> [[TMP3]], <16 x i32>* bitcast ([16 x i32]* @c32 to <16 x i32>*), align 4
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 0 ), align 4
+  %a1  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 1 ), align 4
+  %a2  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 2 ), align 4
+  %a3  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 3 ), align 4
+  %a4  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4 ), align 4
+  %a5  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 5 ), align 4
+  %a6  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 6 ), align 4
+  %a7  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 7 ), align 4
+  %a8  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8 ), align 4
+  %a9  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 9 ), align 4
+  %a10 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 10), align 4
+  %a11 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 11), align 4
+  %a12 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12), align 4
+  %a13 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 13), align 4
+  %a14 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 14), align 4
+  %a15 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 15), align 4
+  %b0  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 0 ), align 4
+  %b1  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 1 ), align 4
+  %b2  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 2 ), align 4
+  %b3  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 3 ), align 4
+  %b4  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4 ), align 4
+  %b5  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 5 ), align 4
+  %b6  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 6 ), align 4
+  %b7  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 7 ), align 4
+  %b8  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8 ), align 4
+  %b9  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 9 ), align 4
+  %b10 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 10), align 4
+  %b11 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 11), align 4
+  %b12 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12), align 4
+  %b13 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 13), align 4
+  %b14 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 14), align 4
+  %b15 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 15), align 4
+  %r0  = call i32 @llvm.smin.i32(i32 %a0 , i32 %b0 )
+  %r1  = call i32 @llvm.smin.i32(i32 %a1 , i32 %b1 )
+  %r2  = call i32 @llvm.smin.i32(i32 %a2 , i32 %b2 )
+  %r3  = call i32 @llvm.smin.i32(i32 %a3 , i32 %b3 )
+  %r4  = call i32 @llvm.smin.i32(i32 %a4 , i32 %b4 )
+  %r5  = call i32 @llvm.smin.i32(i32 %a5 , i32 %b5 )
+  %r6  = call i32 @llvm.smin.i32(i32 %a6 , i32 %b6 )
+  %r7  = call i32 @llvm.smin.i32(i32 %a7 , i32 %b7 )
+  %r8  = call i32 @llvm.smin.i32(i32 %a8 , i32 %b8 )
+  %r9  = call i32 @llvm.smin.i32(i32 %a9 , i32 %b9 )
+  %r10 = call i32 @llvm.smin.i32(i32 %a10, i32 %b10)
+  %r11 = call i32 @llvm.smin.i32(i32 %a11, i32 %b11)
+  %r12 = call i32 @llvm.smin.i32(i32 %a12, i32 %b12)
+  %r13 = call i32 @llvm.smin.i32(i32 %a13, i32 %b13)
+  %r14 = call i32 @llvm.smin.i32(i32 %a14, i32 %b14)
+  %r15 = call i32 @llvm.smin.i32(i32 %a15, i32 %b15)
+  store i32 %r0 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 0 ), align 4
+  store i32 %r1 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 1 ), align 4
+  store i32 %r2 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 2 ), align 4
+  store i32 %r3 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 3 ), align 4
+  store i32 %r4 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4 ), align 4
+  store i32 %r5 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 5 ), align 4
+  store i32 %r6 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 6 ), align 4
+  store i32 %r7 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 7 ), align 4
+  store i32 %r8 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8 ), align 4
+  store i32 %r9 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 9 ), align 4
+  store i32 %r10, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 10), align 4
+  store i32 %r11, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 11), align 4
+  store i32 %r12, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12), align 4
+  store i32 %r13, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 13), align 4
+  store i32 %r14, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 14), align 4
+  store i32 %r15, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 15), align 4
+  ret void
+}
+
+define void @smin_v32i16() {
+; SSE-LABEL: @smin_v32i16(
+; SSE-NEXT:    [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @a16 to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP3:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP5:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @b16 to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP6:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP7:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP8:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP9:%.*]] = call <8 x i16> @llvm.smin.v8i16(<8 x i16> [[TMP1]], <8 x i16> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <8 x i16> @llvm.smin.v8i16(<8 x i16> [[TMP2]], <8 x i16> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <8 x i16> @llvm.smin.v8i16(<8 x i16> [[TMP3]], <8 x i16> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <8 x i16> @llvm.smin.v8i16(<8 x i16> [[TMP4]], <8 x i16> [[TMP8]])
+; SSE-NEXT:    store <8 x i16> [[TMP9]], <8 x i16>* bitcast ([32 x i16]* @c16 to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP10]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP11]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP12]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @smin_v32i16(
+; SLM-NEXT:    [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @a16 to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP3:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP5:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @b16 to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP6:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP7:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP8:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP9:%.*]] = call <8 x i16> @llvm.smin.v8i16(<8 x i16> [[TMP1]], <8 x i16> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <8 x i16> @llvm.smin.v8i16(<8 x i16> [[TMP2]], <8 x i16> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <8 x i16> @llvm.smin.v8i16(<8 x i16> [[TMP3]], <8 x i16> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <8 x i16> @llvm.smin.v8i16(<8 x i16> [[TMP4]], <8 x i16> [[TMP8]])
+; SLM-NEXT:    store <8 x i16> [[TMP9]], <8 x i16>* bitcast ([32 x i16]* @c16 to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP10]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP11]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP12]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @smin_v32i16(
+; AVX-NEXT:    [[TMP1:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([32 x i16]* @a16 to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP2:%.*]] = load <16 x i16>, <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP3:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([32 x i16]* @b16 to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP4:%.*]] = load <16 x i16>, <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP5:%.*]] = call <16 x i16> @llvm.smin.v16i16(<16 x i16> [[TMP1]], <16 x i16> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <16 x i16> @llvm.smin.v16i16(<16 x i16> [[TMP2]], <16 x i16> [[TMP4]])
+; AVX-NEXT:    store <16 x i16> [[TMP5]], <16 x i16>* bitcast ([32 x i16]* @c16 to <16 x i16>*), align 2
+; AVX-NEXT:    store <16 x i16> [[TMP6]], <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @smin_v32i16(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <32 x i16>, <32 x i16>* bitcast ([32 x i16]* @a16 to <32 x i16>*), align 2
+; AVX512-NEXT:    [[TMP2:%.*]] = load <32 x i16>, <32 x i16>* bitcast ([32 x i16]* @b16 to <32 x i16>*), align 2
+; AVX512-NEXT:    [[TMP3:%.*]] = call <32 x i16> @llvm.smin.v32i16(<32 x i16> [[TMP1]], <32 x i16> [[TMP2]])
+; AVX512-NEXT:    store <32 x i16> [[TMP3]], <32 x i16>* bitcast ([32 x i16]* @c16 to <32 x i16>*), align 2
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 0 ), align 2
+  %a1  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 1 ), align 2
+  %a2  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 2 ), align 2
+  %a3  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 3 ), align 2
+  %a4  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 4 ), align 2
+  %a5  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 5 ), align 2
+  %a6  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 6 ), align 2
+  %a7  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 7 ), align 2
+  %a8  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8 ), align 2
+  %a9  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 9 ), align 2
+  %a10 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 10), align 2
+  %a11 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 11), align 2
+  %a12 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 12), align 2
+  %a13 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 13), align 2
+  %a14 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 14), align 2
+  %a15 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 15), align 2
+  %a16 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16), align 2
+  %a17 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 17), align 2
+  %a18 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 18), align 2
+  %a19 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 19), align 2
+  %a20 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 20), align 2
+  %a21 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 21), align 2
+  %a22 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 22), align 2
+  %a23 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 23), align 2
+  %a24 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24), align 2
+  %a25 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 25), align 2
+  %a26 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 26), align 2
+  %a27 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 27), align 2
+  %a28 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 28), align 2
+  %a29 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 29), align 2
+  %a30 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 30), align 2
+  %a31 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 31), align 2
+  %b0  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 0 ), align 2
+  %b1  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 1 ), align 2
+  %b2  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 2 ), align 2
+  %b3  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 3 ), align 2
+  %b4  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 4 ), align 2
+  %b5  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 5 ), align 2
+  %b6  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 6 ), align 2
+  %b7  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 7 ), align 2
+  %b8  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8 ), align 2
+  %b9  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 9 ), align 2
+  %b10 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 10), align 2
+  %b11 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 11), align 2
+  %b12 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 12), align 2
+  %b13 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 13), align 2
+  %b14 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 14), align 2
+  %b15 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 15), align 2
+  %b16 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16), align 2
+  %b17 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 17), align 2
+  %b18 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 18), align 2
+  %b19 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 19), align 2
+  %b20 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 20), align 2
+  %b21 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 21), align 2
+  %b22 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 22), align 2
+  %b23 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 23), align 2
+  %b24 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24), align 2
+  %b25 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 25), align 2
+  %b26 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 26), align 2
+  %b27 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 27), align 2
+  %b28 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 28), align 2
+  %b29 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 29), align 2
+  %b30 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 30), align 2
+  %b31 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 31), align 2
+  %r0  = call i16 @llvm.smin.i16(i16 %a0 , i16 %b0 )
+  %r1  = call i16 @llvm.smin.i16(i16 %a1 , i16 %b1 )
+  %r2  = call i16 @llvm.smin.i16(i16 %a2 , i16 %b2 )
+  %r3  = call i16 @llvm.smin.i16(i16 %a3 , i16 %b3 )
+  %r4  = call i16 @llvm.smin.i16(i16 %a4 , i16 %b4 )
+  %r5  = call i16 @llvm.smin.i16(i16 %a5 , i16 %b5 )
+  %r6  = call i16 @llvm.smin.i16(i16 %a6 , i16 %b6 )
+  %r7  = call i16 @llvm.smin.i16(i16 %a7 , i16 %b7 )
+  %r8  = call i16 @llvm.smin.i16(i16 %a8 , i16 %b8 )
+  %r9  = call i16 @llvm.smin.i16(i16 %a9 , i16 %b9 )
+  %r10 = call i16 @llvm.smin.i16(i16 %a10, i16 %b10)
+  %r11 = call i16 @llvm.smin.i16(i16 %a11, i16 %b11)
+  %r12 = call i16 @llvm.smin.i16(i16 %a12, i16 %b12)
+  %r13 = call i16 @llvm.smin.i16(i16 %a13, i16 %b13)
+  %r14 = call i16 @llvm.smin.i16(i16 %a14, i16 %b14)
+  %r15 = call i16 @llvm.smin.i16(i16 %a15, i16 %b15)
+  %r16 = call i16 @llvm.smin.i16(i16 %a16, i16 %b16)
+  %r17 = call i16 @llvm.smin.i16(i16 %a17, i16 %b17)
+  %r18 = call i16 @llvm.smin.i16(i16 %a18, i16 %b18)
+  %r19 = call i16 @llvm.smin.i16(i16 %a19, i16 %b19)
+  %r20 = call i16 @llvm.smin.i16(i16 %a20, i16 %b20)
+  %r21 = call i16 @llvm.smin.i16(i16 %a21, i16 %b21)
+  %r22 = call i16 @llvm.smin.i16(i16 %a22, i16 %b22)
+  %r23 = call i16 @llvm.smin.i16(i16 %a23, i16 %b23)
+  %r24 = call i16 @llvm.smin.i16(i16 %a24, i16 %b24)
+  %r25 = call i16 @llvm.smin.i16(i16 %a25, i16 %b25)
+  %r26 = call i16 @llvm.smin.i16(i16 %a26, i16 %b26)
+  %r27 = call i16 @llvm.smin.i16(i16 %a27, i16 %b27)
+  %r28 = call i16 @llvm.smin.i16(i16 %a28, i16 %b28)
+  %r29 = call i16 @llvm.smin.i16(i16 %a29, i16 %b29)
+  %r30 = call i16 @llvm.smin.i16(i16 %a30, i16 %b30)
+  %r31 = call i16 @llvm.smin.i16(i16 %a31, i16 %b31)
+  store i16 %r0 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 0 ), align 2
+  store i16 %r1 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 1 ), align 2
+  store i16 %r2 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 2 ), align 2
+  store i16 %r3 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 3 ), align 2
+  store i16 %r4 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 4 ), align 2
+  store i16 %r5 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 5 ), align 2
+  store i16 %r6 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 6 ), align 2
+  store i16 %r7 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 7 ), align 2
+  store i16 %r8 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8 ), align 2
+  store i16 %r9 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 9 ), align 2
+  store i16 %r10, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 10), align 2
+  store i16 %r11, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 11), align 2
+  store i16 %r12, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 12), align 2
+  store i16 %r13, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 13), align 2
+  store i16 %r14, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 14), align 2
+  store i16 %r15, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 15), align 2
+  store i16 %r16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16), align 2
+  store i16 %r17, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 17), align 2
+  store i16 %r18, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 18), align 2
+  store i16 %r19, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 19), align 2
+  store i16 %r20, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 20), align 2
+  store i16 %r21, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 21), align 2
+  store i16 %r22, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 22), align 2
+  store i16 %r23, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 23), align 2
+  store i16 %r24, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24), align 2
+  store i16 %r25, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 25), align 2
+  store i16 %r26, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 26), align 2
+  store i16 %r27, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 27), align 2
+  store i16 %r28, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 28), align 2
+  store i16 %r29, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 29), align 2
+  store i16 %r30, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 30), align 2
+  store i16 %r31, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 31), align 2
+  ret void
+}
+
+define void @smin_v64i8() {
+; SSE-LABEL: @smin_v64i8(
+; SSE-NEXT:    [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @a8 to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP5:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @b8 to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP6:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP7:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP8:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP9:%.*]] = call <16 x i8> @llvm.smin.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <16 x i8> @llvm.smin.v16i8(<16 x i8> [[TMP2]], <16 x i8> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <16 x i8> @llvm.smin.v16i8(<16 x i8> [[TMP3]], <16 x i8> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <16 x i8> @llvm.smin.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP8]])
+; SSE-NEXT:    store <16 x i8> [[TMP9]], <16 x i8>* bitcast ([64 x i8]* @c8 to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP10]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP11]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP12]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @smin_v64i8(
+; SLM-NEXT:    [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @a8 to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP5:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @b8 to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP6:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP7:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP8:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP9:%.*]] = call <16 x i8> @llvm.smin.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <16 x i8> @llvm.smin.v16i8(<16 x i8> [[TMP2]], <16 x i8> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <16 x i8> @llvm.smin.v16i8(<16 x i8> [[TMP3]], <16 x i8> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <16 x i8> @llvm.smin.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP8]])
+; SLM-NEXT:    store <16 x i8> [[TMP9]], <16 x i8>* bitcast ([64 x i8]* @c8 to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP10]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP11]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP12]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @smin_v64i8(
+; AVX-NEXT:    [[TMP1:%.*]] = load <32 x i8>, <32 x i8>* bitcast ([64 x i8]* @a8 to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP2:%.*]] = load <32 x i8>, <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP3:%.*]] = load <32 x i8>, <32 x i8>* bitcast ([64 x i8]* @b8 to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP4:%.*]] = load <32 x i8>, <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP5:%.*]] = call <32 x i8> @llvm.smin.v32i8(<32 x i8> [[TMP1]], <32 x i8> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <32 x i8> @llvm.smin.v32i8(<32 x i8> [[TMP2]], <32 x i8> [[TMP4]])
+; AVX-NEXT:    store <32 x i8> [[TMP5]], <32 x i8>* bitcast ([64 x i8]* @c8 to <32 x i8>*), align 1
+; AVX-NEXT:    store <32 x i8> [[TMP6]], <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @smin_v64i8(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <64 x i8>, <64 x i8>* bitcast ([64 x i8]* @a8 to <64 x i8>*), align 1
+; AVX512-NEXT:    [[TMP2:%.*]] = load <64 x i8>, <64 x i8>* bitcast ([64 x i8]* @b8 to <64 x i8>*), align 1
+; AVX512-NEXT:    [[TMP3:%.*]] = call <64 x i8> @llvm.smin.v64i8(<64 x i8> [[TMP1]], <64 x i8> [[TMP2]])
+; AVX512-NEXT:    store <64 x i8> [[TMP3]], <64 x i8>* bitcast ([64 x i8]* @c8 to <64 x i8>*), align 1
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 0 ), align 1
+  %a1  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 1 ), align 1
+  %a2  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 2 ), align 1
+  %a3  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 3 ), align 1
+  %a4  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 4 ), align 1
+  %a5  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 5 ), align 1
+  %a6  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 6 ), align 1
+  %a7  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 7 ), align 1
+  %a8  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 8 ), align 1
+  %a9  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 9 ), align 1
+  %a10 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 10), align 1
+  %a11 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 11), align 1
+  %a12 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 12), align 1
+  %a13 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 13), align 1
+  %a14 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 14), align 1
+  %a15 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 15), align 1
+  %a16 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16), align 1
+  %a17 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 17), align 1
+  %a18 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 18), align 1
+  %a19 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 19), align 1
+  %a20 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 20), align 1
+  %a21 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 21), align 1
+  %a22 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 22), align 1
+  %a23 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 23), align 1
+  %a24 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 24), align 1
+  %a25 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 25), align 1
+  %a26 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 26), align 1
+  %a27 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 27), align 1
+  %a28 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 28), align 1
+  %a29 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 29), align 1
+  %a30 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 30), align 1
+  %a31 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 31), align 1
+  %a32 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32), align 1
+  %a33 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 33), align 1
+  %a34 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 34), align 1
+  %a35 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 35), align 1
+  %a36 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 36), align 1
+  %a37 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 37), align 1
+  %a38 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 38), align 1
+  %a39 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 39), align 1
+  %a40 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 40), align 1
+  %a41 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 41), align 1
+  %a42 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 42), align 1
+  %a43 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 43), align 1
+  %a44 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 44), align 1
+  %a45 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 45), align 1
+  %a46 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 46), align 1
+  %a47 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 47), align 1
+  %a48 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48), align 1
+  %a49 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 49), align 1
+  %a50 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 50), align 1
+  %a51 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 51), align 1
+  %a52 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 52), align 1
+  %a53 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 53), align 1
+  %a54 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 54), align 1
+  %a55 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 55), align 1
+  %a56 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 56), align 1
+  %a57 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 57), align 1
+  %a58 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 58), align 1
+  %a59 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 59), align 1
+  %a60 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 60), align 1
+  %a61 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 61), align 1
+  %a62 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 62), align 1
+  %a63 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 63), align 1
+  %b0  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 0 ), align 1
+  %b1  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 1 ), align 1
+  %b2  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 2 ), align 1
+  %b3  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 3 ), align 1
+  %b4  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 4 ), align 1
+  %b5  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 5 ), align 1
+  %b6  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 6 ), align 1
+  %b7  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 7 ), align 1
+  %b8  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 8 ), align 1
+  %b9  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 9 ), align 1
+  %b10 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 10), align 1
+  %b11 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 11), align 1
+  %b12 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 12), align 1
+  %b13 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 13), align 1
+  %b14 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 14), align 1
+  %b15 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 15), align 1
+  %b16 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16), align 1
+  %b17 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 17), align 1
+  %b18 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 18), align 1
+  %b19 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 19), align 1
+  %b20 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 20), align 1
+  %b21 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 21), align 1
+  %b22 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 22), align 1
+  %b23 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 23), align 1
+  %b24 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 24), align 1
+  %b25 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 25), align 1
+  %b26 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 26), align 1
+  %b27 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 27), align 1
+  %b28 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 28), align 1
+  %b29 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 29), align 1
+  %b30 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 30), align 1
+  %b31 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 31), align 1
+  %b32 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32), align 1
+  %b33 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 33), align 1
+  %b34 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 34), align 1
+  %b35 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 35), align 1
+  %b36 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 36), align 1
+  %b37 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 37), align 1
+  %b38 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 38), align 1
+  %b39 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 39), align 1
+  %b40 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 40), align 1
+  %b41 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 41), align 1
+  %b42 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 42), align 1
+  %b43 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 43), align 1
+  %b44 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 44), align 1
+  %b45 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 45), align 1
+  %b46 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 46), align 1
+  %b47 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 47), align 1
+  %b48 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48), align 1
+  %b49 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 49), align 1
+  %b50 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 50), align 1
+  %b51 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 51), align 1
+  %b52 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 52), align 1
+  %b53 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 53), align 1
+  %b54 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 54), align 1
+  %b55 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 55), align 1
+  %b56 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 56), align 1
+  %b57 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 57), align 1
+  %b58 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 58), align 1
+  %b59 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 59), align 1
+  %b60 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 60), align 1
+  %b61 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 61), align 1
+  %b62 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 62), align 1
+  %b63 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 63), align 1
+  %r0  = call i8 @llvm.smin.i8(i8 %a0 , i8 %b0 )
+  %r1  = call i8 @llvm.smin.i8(i8 %a1 , i8 %b1 )
+  %r2  = call i8 @llvm.smin.i8(i8 %a2 , i8 %b2 )
+  %r3  = call i8 @llvm.smin.i8(i8 %a3 , i8 %b3 )
+  %r4  = call i8 @llvm.smin.i8(i8 %a4 , i8 %b4 )
+  %r5  = call i8 @llvm.smin.i8(i8 %a5 , i8 %b5 )
+  %r6  = call i8 @llvm.smin.i8(i8 %a6 , i8 %b6 )
+  %r7  = call i8 @llvm.smin.i8(i8 %a7 , i8 %b7 )
+  %r8  = call i8 @llvm.smin.i8(i8 %a8 , i8 %b8 )
+  %r9  = call i8 @llvm.smin.i8(i8 %a9 , i8 %b9 )
+  %r10 = call i8 @llvm.smin.i8(i8 %a10, i8 %b10)
+  %r11 = call i8 @llvm.smin.i8(i8 %a11, i8 %b11)
+  %r12 = call i8 @llvm.smin.i8(i8 %a12, i8 %b12)
+  %r13 = call i8 @llvm.smin.i8(i8 %a13, i8 %b13)
+  %r14 = call i8 @llvm.smin.i8(i8 %a14, i8 %b14)
+  %r15 = call i8 @llvm.smin.i8(i8 %a15, i8 %b15)
+  %r16 = call i8 @llvm.smin.i8(i8 %a16, i8 %b16)
+  %r17 = call i8 @llvm.smin.i8(i8 %a17, i8 %b17)
+  %r18 = call i8 @llvm.smin.i8(i8 %a18, i8 %b18)
+  %r19 = call i8 @llvm.smin.i8(i8 %a19, i8 %b19)
+  %r20 = call i8 @llvm.smin.i8(i8 %a20, i8 %b20)
+  %r21 = call i8 @llvm.smin.i8(i8 %a21, i8 %b21)
+  %r22 = call i8 @llvm.smin.i8(i8 %a22, i8 %b22)
+  %r23 = call i8 @llvm.smin.i8(i8 %a23, i8 %b23)
+  %r24 = call i8 @llvm.smin.i8(i8 %a24, i8 %b24)
+  %r25 = call i8 @llvm.smin.i8(i8 %a25, i8 %b25)
+  %r26 = call i8 @llvm.smin.i8(i8 %a26, i8 %b26)
+  %r27 = call i8 @llvm.smin.i8(i8 %a27, i8 %b27)
+  %r28 = call i8 @llvm.smin.i8(i8 %a28, i8 %b28)
+  %r29 = call i8 @llvm.smin.i8(i8 %a29, i8 %b29)
+  %r30 = call i8 @llvm.smin.i8(i8 %a30, i8 %b30)
+  %r31 = call i8 @llvm.smin.i8(i8 %a31, i8 %b31)
+  %r32 = call i8 @llvm.smin.i8(i8 %a32, i8 %b32)
+  %r33 = call i8 @llvm.smin.i8(i8 %a33, i8 %b33)
+  %r34 = call i8 @llvm.smin.i8(i8 %a34, i8 %b34)
+  %r35 = call i8 @llvm.smin.i8(i8 %a35, i8 %b35)
+  %r36 = call i8 @llvm.smin.i8(i8 %a36, i8 %b36)
+  %r37 = call i8 @llvm.smin.i8(i8 %a37, i8 %b37)
+  %r38 = call i8 @llvm.smin.i8(i8 %a38, i8 %b38)
+  %r39 = call i8 @llvm.smin.i8(i8 %a39, i8 %b39)
+  %r40 = call i8 @llvm.smin.i8(i8 %a40, i8 %b40)
+  %r41 = call i8 @llvm.smin.i8(i8 %a41, i8 %b41)
+  %r42 = call i8 @llvm.smin.i8(i8 %a42, i8 %b42)
+  %r43 = call i8 @llvm.smin.i8(i8 %a43, i8 %b43)
+  %r44 = call i8 @llvm.smin.i8(i8 %a44, i8 %b44)
+  %r45 = call i8 @llvm.smin.i8(i8 %a45, i8 %b45)
+  %r46 = call i8 @llvm.smin.i8(i8 %a46, i8 %b46)
+  %r47 = call i8 @llvm.smin.i8(i8 %a47, i8 %b47)
+  %r48 = call i8 @llvm.smin.i8(i8 %a48, i8 %b48)
+  %r49 = call i8 @llvm.smin.i8(i8 %a49, i8 %b49)
+  %r50 = call i8 @llvm.smin.i8(i8 %a50, i8 %b50)
+  %r51 = call i8 @llvm.smin.i8(i8 %a51, i8 %b51)
+  %r52 = call i8 @llvm.smin.i8(i8 %a52, i8 %b52)
+  %r53 = call i8 @llvm.smin.i8(i8 %a53, i8 %b53)
+  %r54 = call i8 @llvm.smin.i8(i8 %a54, i8 %b54)
+  %r55 = call i8 @llvm.smin.i8(i8 %a55, i8 %b55)
+  %r56 = call i8 @llvm.smin.i8(i8 %a56, i8 %b56)
+  %r57 = call i8 @llvm.smin.i8(i8 %a57, i8 %b57)
+  %r58 = call i8 @llvm.smin.i8(i8 %a58, i8 %b58)
+  %r59 = call i8 @llvm.smin.i8(i8 %a59, i8 %b59)
+  %r60 = call i8 @llvm.smin.i8(i8 %a60, i8 %b60)
+  %r61 = call i8 @llvm.smin.i8(i8 %a61, i8 %b61)
+  %r62 = call i8 @llvm.smin.i8(i8 %a62, i8 %b62)
+  %r63 = call i8 @llvm.smin.i8(i8 %a63, i8 %b63)
+  store i8 %r0 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 0 ), align 1
+  store i8 %r1 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 1 ), align 1
+  store i8 %r2 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 2 ), align 1
+  store i8 %r3 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 3 ), align 1
+  store i8 %r4 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 4 ), align 1
+  store i8 %r5 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 5 ), align 1
+  store i8 %r6 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 6 ), align 1
+  store i8 %r7 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 7 ), align 1
+  store i8 %r8 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 8 ), align 1
+  store i8 %r9 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 9 ), align 1
+  store i8 %r10, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 10), align 1
+  store i8 %r11, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 11), align 1
+  store i8 %r12, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 12), align 1
+  store i8 %r13, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 13), align 1
+  store i8 %r14, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 14), align 1
+  store i8 %r15, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 15), align 1
+  store i8 %r16, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16), align 1
+  store i8 %r17, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 17), align 1
+  store i8 %r18, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 18), align 1
+  store i8 %r19, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 19), align 1
+  store i8 %r20, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 20), align 1
+  store i8 %r21, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 21), align 1
+  store i8 %r22, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 22), align 1
+  store i8 %r23, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 23), align 1
+  store i8 %r24, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 24), align 1
+  store i8 %r25, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 25), align 1
+  store i8 %r26, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 26), align 1
+  store i8 %r27, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 27), align 1
+  store i8 %r28, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 28), align 1
+  store i8 %r29, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 29), align 1
+  store i8 %r30, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 30), align 1
+  store i8 %r31, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 31), align 1
+  store i8 %r32, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32), align 1
+  store i8 %r33, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 33), align 1
+  store i8 %r34, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 34), align 1
+  store i8 %r35, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 35), align 1
+  store i8 %r36, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 36), align 1
+  store i8 %r37, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 37), align 1
+  store i8 %r38, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 38), align 1
+  store i8 %r39, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 39), align 1
+  store i8 %r40, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 40), align 1
+  store i8 %r41, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 41), align 1
+  store i8 %r42, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 42), align 1
+  store i8 %r43, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 43), align 1
+  store i8 %r44, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 44), align 1
+  store i8 %r45, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 45), align 1
+  store i8 %r46, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 46), align 1
+  store i8 %r47, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 47), align 1
+  store i8 %r48, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48), align 1
+  store i8 %r49, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 49), align 1
+  store i8 %r50, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 50), align 1
+  store i8 %r51, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 51), align 1
+  store i8 %r52, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 52), align 1
+  store i8 %r53, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 53), align 1
+  store i8 %r54, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 54), align 1
+  store i8 %r55, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 55), align 1
+  store i8 %r56, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 56), align 1
+  store i8 %r57, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 57), align 1
+  store i8 %r58, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 58), align 1
+  store i8 %r59, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 59), align 1
+  store i8 %r60, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 60), align 1
+  store i8 %r61, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 61), align 1
+  store i8 %r62, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 62), align 1
+  store i8 %r63, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 63), align 1
+  ret void
+}
diff --git a/test/Transforms/SLPVectorizer/X86/arith-umax.ll b/test/Transforms/SLPVectorizer/X86/arith-umax.ll
new file mode 100644
index 00000000000..b3c5e8ac1e3
--- /dev/null
+++ b/test/Transforms/SLPVectorizer/X86/arith-umax.ll
@@ -0,0 +1,762 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -mtriple=x86_64-unknown -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,SSE
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=slm -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,SLM
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=corei7-avx -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=knl -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512F
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=-prefer-256-bit -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512BW
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=+prefer-256-bit -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX256BW
+
+@a64 = common global [8 x i64] zeroinitializer, align 64
+@b64 = common global [8 x i64] zeroinitializer, align 64
+@c64 = common global [8 x i64] zeroinitializer, align 64
+@a32 = common global [16 x i32] zeroinitializer, align 64
+@b32 = common global [16 x i32] zeroinitializer, align 64
+@c32 = common global [16 x i32] zeroinitializer, align 64
+@a16 = common global [32 x i16] zeroinitializer, align 64
+@b16 = common global [32 x i16] zeroinitializer, align 64
+@c16 = common global [32 x i16] zeroinitializer, align 64
+@a8  = common global [64 x i8] zeroinitializer, align 64
+@b8  = common global [64 x i8] zeroinitializer, align 64
+@c8  = common global [64 x i8] zeroinitializer, align 64
+
+declare i64 @llvm.umax.i64(i64, i64)
+declare i32 @llvm.umax.i32(i32, i32)
+declare i16 @llvm.umax.i16(i16, i16)
+declare i8  @llvm.umax.i8 (i8 , i8 )
+
+define void @umax_v8i64() {
+; SSE-LABEL: @umax_v8i64(
+; SSE-NEXT:    [[A0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8
+; SSE-NEXT:    [[A1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8
+; SSE-NEXT:    [[A2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8
+; SSE-NEXT:    [[A3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8
+; SSE-NEXT:    [[A4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8
+; SSE-NEXT:    [[A5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8
+; SSE-NEXT:    [[A6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8
+; SSE-NEXT:    [[A7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8
+; SSE-NEXT:    [[B0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8
+; SSE-NEXT:    [[B1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8
+; SSE-NEXT:    [[B2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8
+; SSE-NEXT:    [[B3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8
+; SSE-NEXT:    [[B4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8
+; SSE-NEXT:    [[B5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8
+; SSE-NEXT:    [[B6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8
+; SSE-NEXT:    [[B7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8
+; SSE-NEXT:    [[R0:%.*]] = call i64 @llvm.umax.i64(i64 [[A0]], i64 [[B0]])
+; SSE-NEXT:    [[R1:%.*]] = call i64 @llvm.umax.i64(i64 [[A1]], i64 [[B1]])
+; SSE-NEXT:    [[R2:%.*]] = call i64 @llvm.umax.i64(i64 [[A2]], i64 [[B2]])
+; SSE-NEXT:    [[R3:%.*]] = call i64 @llvm.umax.i64(i64 [[A3]], i64 [[B3]])
+; SSE-NEXT:    [[R4:%.*]] = call i64 @llvm.umax.i64(i64 [[A4]], i64 [[B4]])
+; SSE-NEXT:    [[R5:%.*]] = call i64 @llvm.umax.i64(i64 [[A5]], i64 [[B5]])
+; SSE-NEXT:    [[R6:%.*]] = call i64 @llvm.umax.i64(i64 [[A6]], i64 [[B6]])
+; SSE-NEXT:    [[R7:%.*]] = call i64 @llvm.umax.i64(i64 [[A7]], i64 [[B7]])
+; SSE-NEXT:    store i64 [[R0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8
+; SSE-NEXT:    store i64 [[R1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8
+; SSE-NEXT:    store i64 [[R2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8
+; SSE-NEXT:    store i64 [[R3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8
+; SSE-NEXT:    store i64 [[R4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8
+; SSE-NEXT:    store i64 [[R5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8
+; SSE-NEXT:    store i64 [[R6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8
+; SSE-NEXT:    store i64 [[R7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @umax_v8i64(
+; SLM-NEXT:    [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @a64 to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP2:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP3:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP4:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP5:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @b64 to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP6:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP7:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP8:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP9:%.*]] = call <2 x i64> @llvm.umax.v2i64(<2 x i64> [[TMP1]], <2 x i64> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <2 x i64> @llvm.umax.v2i64(<2 x i64> [[TMP2]], <2 x i64> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <2 x i64> @llvm.umax.v2i64(<2 x i64> [[TMP3]], <2 x i64> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <2 x i64> @llvm.umax.v2i64(<2 x i64> [[TMP4]], <2 x i64> [[TMP8]])
+; SLM-NEXT:    store <2 x i64> [[TMP9]], <2 x i64>* bitcast ([8 x i64]* @c64 to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP10]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP11]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP12]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @umax_v8i64(
+; AVX-NEXT:    [[TMP1:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @a64 to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP2:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP3:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @b64 to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP4:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP5:%.*]] = call <4 x i64> @llvm.umax.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <4 x i64> @llvm.umax.v4i64(<4 x i64> [[TMP2]], <4 x i64> [[TMP4]])
+; AVX-NEXT:    store <4 x i64> [[TMP5]], <4 x i64>* bitcast ([8 x i64]* @c64 to <4 x i64>*), align 8
+; AVX-NEXT:    store <4 x i64> [[TMP6]], <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @umax_v8i64(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <8 x i64>, <8 x i64>* bitcast ([8 x i64]* @a64 to <8 x i64>*), align 8
+; AVX512-NEXT:    [[TMP2:%.*]] = load <8 x i64>, <8 x i64>* bitcast ([8 x i64]* @b64 to <8 x i64>*), align 8
+; AVX512-NEXT:    [[TMP3:%.*]] = call <8 x i64> @llvm.umax.v8i64(<8 x i64> [[TMP1]], <8 x i64> [[TMP2]])
+; AVX512-NEXT:    store <8 x i64> [[TMP3]], <8 x i64>* bitcast ([8 x i64]* @c64 to <8 x i64>*), align 8
+; AVX512-NEXT:    ret void
+;
+  %a0 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8
+  %a1 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8
+  %a2 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8
+  %a3 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8
+  %a4 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8
+  %a5 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8
+  %a6 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8
+  %a7 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8
+  %b0 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8
+  %b1 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8
+  %b2 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8
+  %b3 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8
+  %b4 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8
+  %b5 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8
+  %b6 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8
+  %b7 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8
+  %r0 = call i64 @llvm.umax.i64(i64 %a0, i64 %b0)
+  %r1 = call i64 @llvm.umax.i64(i64 %a1, i64 %b1)
+  %r2 = call i64 @llvm.umax.i64(i64 %a2, i64 %b2)
+  %r3 = call i64 @llvm.umax.i64(i64 %a3, i64 %b3)
+  %r4 = call i64 @llvm.umax.i64(i64 %a4, i64 %b4)
+  %r5 = call i64 @llvm.umax.i64(i64 %a5, i64 %b5)
+  %r6 = call i64 @llvm.umax.i64(i64 %a6, i64 %b6)
+  %r7 = call i64 @llvm.umax.i64(i64 %a7, i64 %b7)
+  store i64 %r0, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8
+  store i64 %r1, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8
+  store i64 %r2, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8
+  store i64 %r3, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8
+  store i64 %r4, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8
+  store i64 %r5, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8
+  store i64 %r6, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8
+  store i64 %r7, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8
+  ret void
+}
+
+define void @umax_v16i32() {
+; SSE-LABEL: @umax_v16i32(
+; SSE-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP5:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @b32 to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP6:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP7:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP8:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP9:%.*]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[TMP2]], <4 x i32> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[TMP3]], <4 x i32> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP8]])
+; SSE-NEXT:    store <4 x i32> [[TMP9]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP10]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP11]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP12]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @umax_v16i32(
+; SLM-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP5:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @b32 to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP6:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP7:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP8:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP9:%.*]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[TMP2]], <4 x i32> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[TMP3]], <4 x i32> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.umax.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP8]])
+; SLM-NEXT:    store <4 x i32> [[TMP9]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP10]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP11]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP12]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @umax_v16i32(
+; AVX-NEXT:    [[TMP1:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([16 x i32]* @a32 to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP3:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([16 x i32]* @b32 to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP4:%.*]] = load <8 x i32>, <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP5:%.*]] = call <8 x i32> @llvm.umax.v8i32(<8 x i32> [[TMP1]], <8 x i32> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <8 x i32> @llvm.umax.v8i32(<8 x i32> [[TMP2]], <8 x i32> [[TMP4]])
+; AVX-NEXT:    store <8 x i32> [[TMP5]], <8 x i32>* bitcast ([16 x i32]* @c32 to <8 x i32>*), align 4
+; AVX-NEXT:    store <8 x i32> [[TMP6]], <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @umax_v16i32(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <16 x i32>, <16 x i32>* bitcast ([16 x i32]* @a32 to <16 x i32>*), align 4
+; AVX512-NEXT:    [[TMP2:%.*]] = load <16 x i32>, <16 x i32>* bitcast ([16 x i32]* @b32 to <16 x i32>*), align 4
+; AVX512-NEXT:    [[TMP3:%.*]] = call <16 x i32> @llvm.umax.v16i32(<16 x i32> [[TMP1]], <16 x i32> [[TMP2]])
+; AVX512-NEXT:    store <16 x i32> [[TMP3]], <16 x i32>* bitcast ([16 x i32]* @c32 to <16 x i32>*), align 4
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 0 ), align 4
+  %a1  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 1 ), align 4
+  %a2  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 2 ), align 4
+  %a3  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 3 ), align 4
+  %a4  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4 ), align 4
+  %a5  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 5 ), align 4
+  %a6  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 6 ), align 4
+  %a7  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 7 ), align 4
+  %a8  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8 ), align 4
+  %a9  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 9 ), align 4
+  %a10 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 10), align 4
+  %a11 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 11), align 4
+  %a12 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12), align 4
+  %a13 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 13), align 4
+  %a14 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 14), align 4
+  %a15 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 15), align 4
+  %b0  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 0 ), align 4
+  %b1  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 1 ), align 4
+  %b2  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 2 ), align 4
+  %b3  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 3 ), align 4
+  %b4  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4 ), align 4
+  %b5  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 5 ), align 4
+  %b6  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 6 ), align 4
+  %b7  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 7 ), align 4
+  %b8  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8 ), align 4
+  %b9  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 9 ), align 4
+  %b10 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 10), align 4
+  %b11 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 11), align 4
+  %b12 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12), align 4
+  %b13 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 13), align 4
+  %b14 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 14), align 4
+  %b15 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 15), align 4
+  %r0  = call i32 @llvm.umax.i32(i32 %a0 , i32 %b0 )
+  %r1  = call i32 @llvm.umax.i32(i32 %a1 , i32 %b1 )
+  %r2  = call i32 @llvm.umax.i32(i32 %a2 , i32 %b2 )
+  %r3  = call i32 @llvm.umax.i32(i32 %a3 , i32 %b3 )
+  %r4  = call i32 @llvm.umax.i32(i32 %a4 , i32 %b4 )
+  %r5  = call i32 @llvm.umax.i32(i32 %a5 , i32 %b5 )
+  %r6  = call i32 @llvm.umax.i32(i32 %a6 , i32 %b6 )
+  %r7  = call i32 @llvm.umax.i32(i32 %a7 , i32 %b7 )
+  %r8  = call i32 @llvm.umax.i32(i32 %a8 , i32 %b8 )
+  %r9  = call i32 @llvm.umax.i32(i32 %a9 , i32 %b9 )
+  %r10 = call i32 @llvm.umax.i32(i32 %a10, i32 %b10)
+  %r11 = call i32 @llvm.umax.i32(i32 %a11, i32 %b11)
+  %r12 = call i32 @llvm.umax.i32(i32 %a12, i32 %b12)
+  %r13 = call i32 @llvm.umax.i32(i32 %a13, i32 %b13)
+  %r14 = call i32 @llvm.umax.i32(i32 %a14, i32 %b14)
+  %r15 = call i32 @llvm.umax.i32(i32 %a15, i32 %b15)
+  store i32 %r0 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 0 ), align 4
+  store i32 %r1 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 1 ), align 4
+  store i32 %r2 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 2 ), align 4
+  store i32 %r3 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 3 ), align 4
+  store i32 %r4 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4 ), align 4
+  store i32 %r5 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 5 ), align 4
+  store i32 %r6 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 6 ), align 4
+  store i32 %r7 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 7 ), align 4
+  store i32 %r8 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8 ), align 4
+  store i32 %r9 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 9 ), align 4
+  store i32 %r10, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 10), align 4
+  store i32 %r11, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 11), align 4
+  store i32 %r12, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12), align 4
+  store i32 %r13, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 13), align 4
+  store i32 %r14, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 14), align 4
+  store i32 %r15, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 15), align 4
+  ret void
+}
+
+define void @umax_v32i16() {
+; SSE-LABEL: @umax_v32i16(
+; SSE-NEXT:    [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @a16 to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP3:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP5:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @b16 to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP6:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP7:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP8:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP9:%.*]] = call <8 x i16> @llvm.umax.v8i16(<8 x i16> [[TMP1]], <8 x i16> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <8 x i16> @llvm.umax.v8i16(<8 x i16> [[TMP2]], <8 x i16> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <8 x i16> @llvm.umax.v8i16(<8 x i16> [[TMP3]], <8 x i16> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <8 x i16> @llvm.umax.v8i16(<8 x i16> [[TMP4]], <8 x i16> [[TMP8]])
+; SSE-NEXT:    store <8 x i16> [[TMP9]], <8 x i16>* bitcast ([32 x i16]* @c16 to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP10]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP11]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP12]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @umax_v32i16(
+; SLM-NEXT:    [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @a16 to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP3:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP5:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @b16 to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP6:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP7:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP8:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP9:%.*]] = call <8 x i16> @llvm.umax.v8i16(<8 x i16> [[TMP1]], <8 x i16> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <8 x i16> @llvm.umax.v8i16(<8 x i16> [[TMP2]], <8 x i16> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <8 x i16> @llvm.umax.v8i16(<8 x i16> [[TMP3]], <8 x i16> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <8 x i16> @llvm.umax.v8i16(<8 x i16> [[TMP4]], <8 x i16> [[TMP8]])
+; SLM-NEXT:    store <8 x i16> [[TMP9]], <8 x i16>* bitcast ([32 x i16]* @c16 to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP10]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP11]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP12]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @umax_v32i16(
+; AVX-NEXT:    [[TMP1:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([32 x i16]* @a16 to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP2:%.*]] = load <16 x i16>, <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP3:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([32 x i16]* @b16 to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP4:%.*]] = load <16 x i16>, <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP5:%.*]] = call <16 x i16> @llvm.umax.v16i16(<16 x i16> [[TMP1]], <16 x i16> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <16 x i16> @llvm.umax.v16i16(<16 x i16> [[TMP2]], <16 x i16> [[TMP4]])
+; AVX-NEXT:    store <16 x i16> [[TMP5]], <16 x i16>* bitcast ([32 x i16]* @c16 to <16 x i16>*), align 2
+; AVX-NEXT:    store <16 x i16> [[TMP6]], <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @umax_v32i16(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <32 x i16>, <32 x i16>* bitcast ([32 x i16]* @a16 to <32 x i16>*), align 2
+; AVX512-NEXT:    [[TMP2:%.*]] = load <32 x i16>, <32 x i16>* bitcast ([32 x i16]* @b16 to <32 x i16>*), align 2
+; AVX512-NEXT:    [[TMP3:%.*]] = call <32 x i16> @llvm.umax.v32i16(<32 x i16> [[TMP1]], <32 x i16> [[TMP2]])
+; AVX512-NEXT:    store <32 x i16> [[TMP3]], <32 x i16>* bitcast ([32 x i16]* @c16 to <32 x i16>*), align 2
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 0 ), align 2
+  %a1  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 1 ), align 2
+  %a2  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 2 ), align 2
+  %a3  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 3 ), align 2
+  %a4  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 4 ), align 2
+  %a5  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 5 ), align 2
+  %a6  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 6 ), align 2
+  %a7  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 7 ), align 2
+  %a8  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8 ), align 2
+  %a9  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 9 ), align 2
+  %a10 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 10), align 2
+  %a11 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 11), align 2
+  %a12 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 12), align 2
+  %a13 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 13), align 2
+  %a14 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 14), align 2
+  %a15 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 15), align 2
+  %a16 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16), align 2
+  %a17 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 17), align 2
+  %a18 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 18), align 2
+  %a19 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 19), align 2
+  %a20 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 20), align 2
+  %a21 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 21), align 2
+  %a22 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 22), align 2
+  %a23 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 23), align 2
+  %a24 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24), align 2
+  %a25 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 25), align 2
+  %a26 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 26), align 2
+  %a27 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 27), align 2
+  %a28 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 28), align 2
+  %a29 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 29), align 2
+  %a30 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 30), align 2
+  %a31 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 31), align 2
+  %b0  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 0 ), align 2
+  %b1  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 1 ), align 2
+  %b2  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 2 ), align 2
+  %b3  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 3 ), align 2
+  %b4  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 4 ), align 2
+  %b5  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 5 ), align 2
+  %b6  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 6 ), align 2
+  %b7  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 7 ), align 2
+  %b8  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8 ), align 2
+  %b9  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 9 ), align 2
+  %b10 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 10), align 2
+  %b11 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 11), align 2
+  %b12 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 12), align 2
+  %b13 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 13), align 2
+  %b14 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 14), align 2
+  %b15 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 15), align 2
+  %b16 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16), align 2
+  %b17 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 17), align 2
+  %b18 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 18), align 2
+  %b19 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 19), align 2
+  %b20 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 20), align 2
+  %b21 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 21), align 2
+  %b22 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 22), align 2
+  %b23 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 23), align 2
+  %b24 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24), align 2
+  %b25 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 25), align 2
+  %b26 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 26), align 2
+  %b27 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 27), align 2
+  %b28 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 28), align 2
+  %b29 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 29), align 2
+  %b30 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 30), align 2
+  %b31 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 31), align 2
+  %r0  = call i16 @llvm.umax.i16(i16 %a0 , i16 %b0 )
+  %r1  = call i16 @llvm.umax.i16(i16 %a1 , i16 %b1 )
+  %r2  = call i16 @llvm.umax.i16(i16 %a2 , i16 %b2 )
+  %r3  = call i16 @llvm.umax.i16(i16 %a3 , i16 %b3 )
+  %r4  = call i16 @llvm.umax.i16(i16 %a4 , i16 %b4 )
+  %r5  = call i16 @llvm.umax.i16(i16 %a5 , i16 %b5 )
+  %r6  = call i16 @llvm.umax.i16(i16 %a6 , i16 %b6 )
+  %r7  = call i16 @llvm.umax.i16(i16 %a7 , i16 %b7 )
+  %r8  = call i16 @llvm.umax.i16(i16 %a8 , i16 %b8 )
+  %r9  = call i16 @llvm.umax.i16(i16 %a9 , i16 %b9 )
+  %r10 = call i16 @llvm.umax.i16(i16 %a10, i16 %b10)
+  %r11 = call i16 @llvm.umax.i16(i16 %a11, i16 %b11)
+  %r12 = call i16 @llvm.umax.i16(i16 %a12, i16 %b12)
+  %r13 = call i16 @llvm.umax.i16(i16 %a13, i16 %b13)
+  %r14 = call i16 @llvm.umax.i16(i16 %a14, i16 %b14)
+  %r15 = call i16 @llvm.umax.i16(i16 %a15, i16 %b15)
+  %r16 = call i16 @llvm.umax.i16(i16 %a16, i16 %b16)
+  %r17 = call i16 @llvm.umax.i16(i16 %a17, i16 %b17)
+  %r18 = call i16 @llvm.umax.i16(i16 %a18, i16 %b18)
+  %r19 = call i16 @llvm.umax.i16(i16 %a19, i16 %b19)
+  %r20 = call i16 @llvm.umax.i16(i16 %a20, i16 %b20)
+  %r21 = call i16 @llvm.umax.i16(i16 %a21, i16 %b21)
+  %r22 = call i16 @llvm.umax.i16(i16 %a22, i16 %b22)
+  %r23 = call i16 @llvm.umax.i16(i16 %a23, i16 %b23)
+  %r24 = call i16 @llvm.umax.i16(i16 %a24, i16 %b24)
+  %r25 = call i16 @llvm.umax.i16(i16 %a25, i16 %b25)
+  %r26 = call i16 @llvm.umax.i16(i16 %a26, i16 %b26)
+  %r27 = call i16 @llvm.umax.i16(i16 %a27, i16 %b27)
+  %r28 = call i16 @llvm.umax.i16(i16 %a28, i16 %b28)
+  %r29 = call i16 @llvm.umax.i16(i16 %a29, i16 %b29)
+  %r30 = call i16 @llvm.umax.i16(i16 %a30, i16 %b30)
+  %r31 = call i16 @llvm.umax.i16(i16 %a31, i16 %b31)
+  store i16 %r0 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 0 ), align 2
+  store i16 %r1 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 1 ), align 2
+  store i16 %r2 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 2 ), align 2
+  store i16 %r3 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 3 ), align 2
+  store i16 %r4 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 4 ), align 2
+  store i16 %r5 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 5 ), align 2
+  store i16 %r6 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 6 ), align 2
+  store i16 %r7 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 7 ), align 2
+  store i16 %r8 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8 ), align 2
+  store i16 %r9 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 9 ), align 2
+  store i16 %r10, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 10), align 2
+  store i16 %r11, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 11), align 2
+  store i16 %r12, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 12), align 2
+  store i16 %r13, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 13), align 2
+  store i16 %r14, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 14), align 2
+  store i16 %r15, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 15), align 2
+  store i16 %r16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16), align 2
+  store i16 %r17, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 17), align 2
+  store i16 %r18, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 18), align 2
+  store i16 %r19, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 19), align 2
+  store i16 %r20, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 20), align 2
+  store i16 %r21, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 21), align 2
+  store i16 %r22, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 22), align 2
+  store i16 %r23, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 23), align 2
+  store i16 %r24, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24), align 2
+  store i16 %r25, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 25), align 2
+  store i16 %r26, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 26), align 2
+  store i16 %r27, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 27), align 2
+  store i16 %r28, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 28), align 2
+  store i16 %r29, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 29), align 2
+  store i16 %r30, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 30), align 2
+  store i16 %r31, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 31), align 2
+  ret void
+}
+
+define void @umax_v64i8() {
+; SSE-LABEL: @umax_v64i8(
+; SSE-NEXT:    [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @a8 to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP5:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @b8 to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP6:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP7:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP8:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP9:%.*]] = call <16 x i8> @llvm.umax.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <16 x i8> @llvm.umax.v16i8(<16 x i8> [[TMP2]], <16 x i8> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <16 x i8> @llvm.umax.v16i8(<16 x i8> [[TMP3]], <16 x i8> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <16 x i8> @llvm.umax.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP8]])
+; SSE-NEXT:    store <16 x i8> [[TMP9]], <16 x i8>* bitcast ([64 x i8]* @c8 to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP10]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP11]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP12]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @umax_v64i8(
+; SLM-NEXT:    [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @a8 to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP5:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @b8 to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP6:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP7:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP8:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP9:%.*]] = call <16 x i8> @llvm.umax.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <16 x i8> @llvm.umax.v16i8(<16 x i8> [[TMP2]], <16 x i8> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <16 x i8> @llvm.umax.v16i8(<16 x i8> [[TMP3]], <16 x i8> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <16 x i8> @llvm.umax.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP8]])
+; SLM-NEXT:    store <16 x i8> [[TMP9]], <16 x i8>* bitcast ([64 x i8]* @c8 to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP10]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP11]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP12]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @umax_v64i8(
+; AVX-NEXT:    [[TMP1:%.*]] = load <32 x i8>, <32 x i8>* bitcast ([64 x i8]* @a8 to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP2:%.*]] = load <32 x i8>, <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP3:%.*]] = load <32 x i8>, <32 x i8>* bitcast ([64 x i8]* @b8 to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP4:%.*]] = load <32 x i8>, <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP5:%.*]] = call <32 x i8> @llvm.umax.v32i8(<32 x i8> [[TMP1]], <32 x i8> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <32 x i8> @llvm.umax.v32i8(<32 x i8> [[TMP2]], <32 x i8> [[TMP4]])
+; AVX-NEXT:    store <32 x i8> [[TMP5]], <32 x i8>* bitcast ([64 x i8]* @c8 to <32 x i8>*), align 1
+; AVX-NEXT:    store <32 x i8> [[TMP6]], <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @umax_v64i8(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <64 x i8>, <64 x i8>* bitcast ([64 x i8]* @a8 to <64 x i8>*), align 1
+; AVX512-NEXT:    [[TMP2:%.*]] = load <64 x i8>, <64 x i8>* bitcast ([64 x i8]* @b8 to <64 x i8>*), align 1
+; AVX512-NEXT:    [[TMP3:%.*]] = call <64 x i8> @llvm.umax.v64i8(<64 x i8> [[TMP1]], <64 x i8> [[TMP2]])
+; AVX512-NEXT:    store <64 x i8> [[TMP3]], <64 x i8>* bitcast ([64 x i8]* @c8 to <64 x i8>*), align 1
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 0 ), align 1
+  %a1  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 1 ), align 1
+  %a2  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 2 ), align 1
+  %a3  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 3 ), align 1
+  %a4  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 4 ), align 1
+  %a5  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 5 ), align 1
+  %a6  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 6 ), align 1
+  %a7  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 7 ), align 1
+  %a8  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 8 ), align 1
+  %a9  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 9 ), align 1
+  %a10 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 10), align 1
+  %a11 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 11), align 1
+  %a12 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 12), align 1
+  %a13 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 13), align 1
+  %a14 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 14), align 1
+  %a15 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 15), align 1
+  %a16 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16), align 1
+  %a17 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 17), align 1
+  %a18 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 18), align 1
+  %a19 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 19), align 1
+  %a20 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 20), align 1
+  %a21 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 21), align 1
+  %a22 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 22), align 1
+  %a23 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 23), align 1
+  %a24 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 24), align 1
+  %a25 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 25), align 1
+  %a26 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 26), align 1
+  %a27 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 27), align 1
+  %a28 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 28), align 1
+  %a29 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 29), align 1
+  %a30 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 30), align 1
+  %a31 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 31), align 1
+  %a32 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32), align 1
+  %a33 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 33), align 1
+  %a34 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 34), align 1
+  %a35 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 35), align 1
+  %a36 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 36), align 1
+  %a37 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 37), align 1
+  %a38 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 38), align 1
+  %a39 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 39), align 1
+  %a40 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 40), align 1
+  %a41 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 41), align 1
+  %a42 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 42), align 1
+  %a43 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 43), align 1
+  %a44 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 44), align 1
+  %a45 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 45), align 1
+  %a46 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 46), align 1
+  %a47 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 47), align 1
+  %a48 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48), align 1
+  %a49 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 49), align 1
+  %a50 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 50), align 1
+  %a51 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 51), align 1
+  %a52 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 52), align 1
+  %a53 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 53), align 1
+  %a54 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 54), align 1
+  %a55 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 55), align 1
+  %a56 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 56), align 1
+  %a57 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 57), align 1
+  %a58 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 58), align 1
+  %a59 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 59), align 1
+  %a60 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 60), align 1
+  %a61 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 61), align 1
+  %a62 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 62), align 1
+  %a63 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 63), align 1
+  %b0  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 0 ), align 1
+  %b1  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 1 ), align 1
+  %b2  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 2 ), align 1
+  %b3  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 3 ), align 1
+  %b4  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 4 ), align 1
+  %b5  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 5 ), align 1
+  %b6  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 6 ), align 1
+  %b7  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 7 ), align 1
+  %b8  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 8 ), align 1
+  %b9  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 9 ), align 1
+  %b10 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 10), align 1
+  %b11 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 11), align 1
+  %b12 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 12), align 1
+  %b13 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 13), align 1
+  %b14 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 14), align 1
+  %b15 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 15), align 1
+  %b16 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16), align 1
+  %b17 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 17), align 1
+  %b18 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 18), align 1
+  %b19 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 19), align 1
+  %b20 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 20), align 1
+  %b21 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 21), align 1
+  %b22 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 22), align 1
+  %b23 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 23), align 1
+  %b24 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 24), align 1
+  %b25 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 25), align 1
+  %b26 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 26), align 1
+  %b27 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 27), align 1
+  %b28 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 28), align 1
+  %b29 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 29), align 1
+  %b30 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 30), align 1
+  %b31 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 31), align 1
+  %b32 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32), align 1
+  %b33 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 33), align 1
+  %b34 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 34), align 1
+  %b35 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 35), align 1
+  %b36 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 36), align 1
+  %b37 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 37), align 1
+  %b38 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 38), align 1
+  %b39 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 39), align 1
+  %b40 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 40), align 1
+  %b41 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 41), align 1
+  %b42 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 42), align 1
+  %b43 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 43), align 1
+  %b44 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 44), align 1
+  %b45 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 45), align 1
+  %b46 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 46), align 1
+  %b47 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 47), align 1
+  %b48 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48), align 1
+  %b49 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 49), align 1
+  %b50 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 50), align 1
+  %b51 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 51), align 1
+  %b52 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 52), align 1
+  %b53 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 53), align 1
+  %b54 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 54), align 1
+  %b55 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 55), align 1
+  %b56 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 56), align 1
+  %b57 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 57), align 1
+  %b58 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 58), align 1
+  %b59 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 59), align 1
+  %b60 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 60), align 1
+  %b61 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 61), align 1
+  %b62 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 62), align 1
+  %b63 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 63), align 1
+  %r0  = call i8 @llvm.umax.i8(i8 %a0 , i8 %b0 )
+  %r1  = call i8 @llvm.umax.i8(i8 %a1 , i8 %b1 )
+  %r2  = call i8 @llvm.umax.i8(i8 %a2 , i8 %b2 )
+  %r3  = call i8 @llvm.umax.i8(i8 %a3 , i8 %b3 )
+  %r4  = call i8 @llvm.umax.i8(i8 %a4 , i8 %b4 )
+  %r5  = call i8 @llvm.umax.i8(i8 %a5 , i8 %b5 )
+  %r6  = call i8 @llvm.umax.i8(i8 %a6 , i8 %b6 )
+  %r7  = call i8 @llvm.umax.i8(i8 %a7 , i8 %b7 )
+  %r8  = call i8 @llvm.umax.i8(i8 %a8 , i8 %b8 )
+  %r9  = call i8 @llvm.umax.i8(i8 %a9 , i8 %b9 )
+  %r10 = call i8 @llvm.umax.i8(i8 %a10, i8 %b10)
+  %r11 = call i8 @llvm.umax.i8(i8 %a11, i8 %b11)
+  %r12 = call i8 @llvm.umax.i8(i8 %a12, i8 %b12)
+  %r13 = call i8 @llvm.umax.i8(i8 %a13, i8 %b13)
+  %r14 = call i8 @llvm.umax.i8(i8 %a14, i8 %b14)
+  %r15 = call i8 @llvm.umax.i8(i8 %a15, i8 %b15)
+  %r16 = call i8 @llvm.umax.i8(i8 %a16, i8 %b16)
+  %r17 = call i8 @llvm.umax.i8(i8 %a17, i8 %b17)
+  %r18 = call i8 @llvm.umax.i8(i8 %a18, i8 %b18)
+  %r19 = call i8 @llvm.umax.i8(i8 %a19, i8 %b19)
+  %r20 = call i8 @llvm.umax.i8(i8 %a20, i8 %b20)
+  %r21 = call i8 @llvm.umax.i8(i8 %a21, i8 %b21)
+  %r22 = call i8 @llvm.umax.i8(i8 %a22, i8 %b22)
+  %r23 = call i8 @llvm.umax.i8(i8 %a23, i8 %b23)
+  %r24 = call i8 @llvm.umax.i8(i8 %a24, i8 %b24)
+  %r25 = call i8 @llvm.umax.i8(i8 %a25, i8 %b25)
+  %r26 = call i8 @llvm.umax.i8(i8 %a26, i8 %b26)
+  %r27 = call i8 @llvm.umax.i8(i8 %a27, i8 %b27)
+  %r28 = call i8 @llvm.umax.i8(i8 %a28, i8 %b28)
+  %r29 = call i8 @llvm.umax.i8(i8 %a29, i8 %b29)
+  %r30 = call i8 @llvm.umax.i8(i8 %a30, i8 %b30)
+  %r31 = call i8 @llvm.umax.i8(i8 %a31, i8 %b31)
+  %r32 = call i8 @llvm.umax.i8(i8 %a32, i8 %b32)
+  %r33 = call i8 @llvm.umax.i8(i8 %a33, i8 %b33)
+  %r34 = call i8 @llvm.umax.i8(i8 %a34, i8 %b34)
+  %r35 = call i8 @llvm.umax.i8(i8 %a35, i8 %b35)
+  %r36 = call i8 @llvm.umax.i8(i8 %a36, i8 %b36)
+  %r37 = call i8 @llvm.umax.i8(i8 %a37, i8 %b37)
+  %r38 = call i8 @llvm.umax.i8(i8 %a38, i8 %b38)
+  %r39 = call i8 @llvm.umax.i8(i8 %a39, i8 %b39)
+  %r40 = call i8 @llvm.umax.i8(i8 %a40, i8 %b40)
+  %r41 = call i8 @llvm.umax.i8(i8 %a41, i8 %b41)
+  %r42 = call i8 @llvm.umax.i8(i8 %a42, i8 %b42)
+  %r43 = call i8 @llvm.umax.i8(i8 %a43, i8 %b43)
+  %r44 = call i8 @llvm.umax.i8(i8 %a44, i8 %b44)
+  %r45 = call i8 @llvm.umax.i8(i8 %a45, i8 %b45)
+  %r46 = call i8 @llvm.umax.i8(i8 %a46, i8 %b46)
+  %r47 = call i8 @llvm.umax.i8(i8 %a47, i8 %b47)
+  %r48 = call i8 @llvm.umax.i8(i8 %a48, i8 %b48)
+  %r49 = call i8 @llvm.umax.i8(i8 %a49, i8 %b49)
+  %r50 = call i8 @llvm.umax.i8(i8 %a50, i8 %b50)
+  %r51 = call i8 @llvm.umax.i8(i8 %a51, i8 %b51)
+  %r52 = call i8 @llvm.umax.i8(i8 %a52, i8 %b52)
+  %r53 = call i8 @llvm.umax.i8(i8 %a53, i8 %b53)
+  %r54 = call i8 @llvm.umax.i8(i8 %a54, i8 %b54)
+  %r55 = call i8 @llvm.umax.i8(i8 %a55, i8 %b55)
+  %r56 = call i8 @llvm.umax.i8(i8 %a56, i8 %b56)
+  %r57 = call i8 @llvm.umax.i8(i8 %a57, i8 %b57)
+  %r58 = call i8 @llvm.umax.i8(i8 %a58, i8 %b58)
+  %r59 = call i8 @llvm.umax.i8(i8 %a59, i8 %b59)
+  %r60 = call i8 @llvm.umax.i8(i8 %a60, i8 %b60)
+  %r61 = call i8 @llvm.umax.i8(i8 %a61, i8 %b61)
+  %r62 = call i8 @llvm.umax.i8(i8 %a62, i8 %b62)
+  %r63 = call i8 @llvm.umax.i8(i8 %a63, i8 %b63)
+  store i8 %r0 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 0 ), align 1
+  store i8 %r1 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 1 ), align 1
+  store i8 %r2 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 2 ), align 1
+  store i8 %r3 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 3 ), align 1
+  store i8 %r4 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 4 ), align 1
+  store i8 %r5 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 5 ), align 1
+  store i8 %r6 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 6 ), align 1
+  store i8 %r7 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 7 ), align 1
+  store i8 %r8 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 8 ), align 1
+  store i8 %r9 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 9 ), align 1
+  store i8 %r10, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 10), align 1
+  store i8 %r11, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 11), align 1
+  store i8 %r12, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 12), align 1
+  store i8 %r13, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 13), align 1
+  store i8 %r14, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 14), align 1
+  store i8 %r15, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 15), align 1
+  store i8 %r16, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16), align 1
+  store i8 %r17, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 17), align 1
+  store i8 %r18, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 18), align 1
+  store i8 %r19, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 19), align 1
+  store i8 %r20, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 20), align 1
+  store i8 %r21, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 21), align 1
+  store i8 %r22, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 22), align 1
+  store i8 %r23, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 23), align 1
+  store i8 %r24, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 24), align 1
+  store i8 %r25, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 25), align 1
+  store i8 %r26, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 26), align 1
+  store i8 %r27, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 27), align 1
+  store i8 %r28, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 28), align 1
+  store i8 %r29, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 29), align 1
+  store i8 %r30, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 30), align 1
+  store i8 %r31, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 31), align 1
+  store i8 %r32, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32), align 1
+  store i8 %r33, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 33), align 1
+  store i8 %r34, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 34), align 1
+  store i8 %r35, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 35), align 1
+  store i8 %r36, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 36), align 1
+  store i8 %r37, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 37), align 1
+  store i8 %r38, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 38), align 1
+  store i8 %r39, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 39), align 1
+  store i8 %r40, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 40), align 1
+  store i8 %r41, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 41), align 1
+  store i8 %r42, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 42), align 1
+  store i8 %r43, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 43), align 1
+  store i8 %r44, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 44), align 1
+  store i8 %r45, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 45), align 1
+  store i8 %r46, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 46), align 1
+  store i8 %r47, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 47), align 1
+  store i8 %r48, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48), align 1
+  store i8 %r49, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 49), align 1
+  store i8 %r50, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 50), align 1
+  store i8 %r51, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 51), align 1
+  store i8 %r52, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 52), align 1
+  store i8 %r53, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 53), align 1
+  store i8 %r54, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 54), align 1
+  store i8 %r55, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 55), align 1
+  store i8 %r56, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 56), align 1
+  store i8 %r57, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 57), align 1
+  store i8 %r58, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 58), align 1
+  store i8 %r59, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 59), align 1
+  store i8 %r60, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 60), align 1
+  store i8 %r61, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 61), align 1
+  store i8 %r62, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 62), align 1
+  store i8 %r63, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 63), align 1
+  ret void
+}
diff --git a/test/Transforms/SLPVectorizer/X86/arith-umin.ll b/test/Transforms/SLPVectorizer/X86/arith-umin.ll
new file mode 100644
index 00000000000..3a501d1ed7a
--- /dev/null
+++ b/test/Transforms/SLPVectorizer/X86/arith-umin.ll
@@ -0,0 +1,762 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -mtriple=x86_64-unknown -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,SSE
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=slm -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,SLM
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=corei7-avx -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=knl -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512F
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=-prefer-256-bit -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512BW
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skx -mattr=+prefer-256-bit -basic-aa -slp-vectorizer -S | FileCheck %s --check-prefixes=CHECK,AVX,AVX256BW
+
+@a64 = common global [8 x i64] zeroinitializer, align 64
+@b64 = common global [8 x i64] zeroinitializer, align 64
+@c64 = common global [8 x i64] zeroinitializer, align 64
+@a32 = common global [16 x i32] zeroinitializer, align 64
+@b32 = common global [16 x i32] zeroinitializer, align 64
+@c32 = common global [16 x i32] zeroinitializer, align 64
+@a16 = common global [32 x i16] zeroinitializer, align 64
+@b16 = common global [32 x i16] zeroinitializer, align 64
+@c16 = common global [32 x i16] zeroinitializer, align 64
+@a8  = common global [64 x i8] zeroinitializer, align 64
+@b8  = common global [64 x i8] zeroinitializer, align 64
+@c8  = common global [64 x i8] zeroinitializer, align 64
+
+declare i64 @llvm.umin.i64(i64, i64)
+declare i32 @llvm.umin.i32(i32, i32)
+declare i16 @llvm.umin.i16(i16, i16)
+declare i8  @llvm.umin.i8 (i8 , i8 )
+
+define void @umin_v8i64() {
+; SSE-LABEL: @umin_v8i64(
+; SSE-NEXT:    [[A0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8
+; SSE-NEXT:    [[A1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8
+; SSE-NEXT:    [[A2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8
+; SSE-NEXT:    [[A3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8
+; SSE-NEXT:    [[A4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8
+; SSE-NEXT:    [[A5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8
+; SSE-NEXT:    [[A6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8
+; SSE-NEXT:    [[A7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8
+; SSE-NEXT:    [[B0:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8
+; SSE-NEXT:    [[B1:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8
+; SSE-NEXT:    [[B2:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8
+; SSE-NEXT:    [[B3:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8
+; SSE-NEXT:    [[B4:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8
+; SSE-NEXT:    [[B5:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8
+; SSE-NEXT:    [[B6:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8
+; SSE-NEXT:    [[B7:%.*]] = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8
+; SSE-NEXT:    [[R0:%.*]] = call i64 @llvm.umin.i64(i64 [[A0]], i64 [[B0]])
+; SSE-NEXT:    [[R1:%.*]] = call i64 @llvm.umin.i64(i64 [[A1]], i64 [[B1]])
+; SSE-NEXT:    [[R2:%.*]] = call i64 @llvm.umin.i64(i64 [[A2]], i64 [[B2]])
+; SSE-NEXT:    [[R3:%.*]] = call i64 @llvm.umin.i64(i64 [[A3]], i64 [[B3]])
+; SSE-NEXT:    [[R4:%.*]] = call i64 @llvm.umin.i64(i64 [[A4]], i64 [[B4]])
+; SSE-NEXT:    [[R5:%.*]] = call i64 @llvm.umin.i64(i64 [[A5]], i64 [[B5]])
+; SSE-NEXT:    [[R6:%.*]] = call i64 @llvm.umin.i64(i64 [[A6]], i64 [[B6]])
+; SSE-NEXT:    [[R7:%.*]] = call i64 @llvm.umin.i64(i64 [[A7]], i64 [[B7]])
+; SSE-NEXT:    store i64 [[R0]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8
+; SSE-NEXT:    store i64 [[R1]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8
+; SSE-NEXT:    store i64 [[R2]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8
+; SSE-NEXT:    store i64 [[R3]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8
+; SSE-NEXT:    store i64 [[R4]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8
+; SSE-NEXT:    store i64 [[R5]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8
+; SSE-NEXT:    store i64 [[R6]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8
+; SSE-NEXT:    store i64 [[R7]], i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @umin_v8i64(
+; SLM-NEXT:    [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @a64 to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP2:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP3:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP4:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP5:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([8 x i64]* @b64 to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP6:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP7:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP8:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    [[TMP9:%.*]] = call <2 x i64> @llvm.umin.v2i64(<2 x i64> [[TMP1]], <2 x i64> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <2 x i64> @llvm.umin.v2i64(<2 x i64> [[TMP2]], <2 x i64> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <2 x i64> @llvm.umin.v2i64(<2 x i64> [[TMP3]], <2 x i64> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <2 x i64> @llvm.umin.v2i64(<2 x i64> [[TMP4]], <2 x i64> [[TMP8]])
+; SLM-NEXT:    store <2 x i64> [[TMP9]], <2 x i64>* bitcast ([8 x i64]* @c64 to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP10]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2) to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP11]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <2 x i64>*), align 8
+; SLM-NEXT:    store <2 x i64> [[TMP12]], <2 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6) to <2 x i64>*), align 8
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @umin_v8i64(
+; AVX-NEXT:    [[TMP1:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @a64 to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP2:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP3:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([8 x i64]* @b64 to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP4:%.*]] = load <4 x i64>, <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    [[TMP5:%.*]] = call <4 x i64> @llvm.umin.v4i64(<4 x i64> [[TMP1]], <4 x i64> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <4 x i64> @llvm.umin.v4i64(<4 x i64> [[TMP2]], <4 x i64> [[TMP4]])
+; AVX-NEXT:    store <4 x i64> [[TMP5]], <4 x i64>* bitcast ([8 x i64]* @c64 to <4 x i64>*), align 8
+; AVX-NEXT:    store <4 x i64> [[TMP6]], <4 x i64>* bitcast (i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4) to <4 x i64>*), align 8
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @umin_v8i64(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <8 x i64>, <8 x i64>* bitcast ([8 x i64]* @a64 to <8 x i64>*), align 8
+; AVX512-NEXT:    [[TMP2:%.*]] = load <8 x i64>, <8 x i64>* bitcast ([8 x i64]* @b64 to <8 x i64>*), align 8
+; AVX512-NEXT:    [[TMP3:%.*]] = call <8 x i64> @llvm.umin.v8i64(<8 x i64> [[TMP1]], <8 x i64> [[TMP2]])
+; AVX512-NEXT:    store <8 x i64> [[TMP3]], <8 x i64>* bitcast ([8 x i64]* @c64 to <8 x i64>*), align 8
+; AVX512-NEXT:    ret void
+;
+  %a0 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 0), align 8
+  %a1 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 1), align 8
+  %a2 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 2), align 8
+  %a3 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 3), align 8
+  %a4 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 4), align 8
+  %a5 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 5), align 8
+  %a6 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 6), align 8
+  %a7 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @a64, i32 0, i64 7), align 8
+  %b0 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 0), align 8
+  %b1 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 1), align 8
+  %b2 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 2), align 8
+  %b3 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 3), align 8
+  %b4 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 4), align 8
+  %b5 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 5), align 8
+  %b6 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 6), align 8
+  %b7 = load i64, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @b64, i32 0, i64 7), align 8
+  %r0 = call i64 @llvm.umin.i64(i64 %a0, i64 %b0)
+  %r1 = call i64 @llvm.umin.i64(i64 %a1, i64 %b1)
+  %r2 = call i64 @llvm.umin.i64(i64 %a2, i64 %b2)
+  %r3 = call i64 @llvm.umin.i64(i64 %a3, i64 %b3)
+  %r4 = call i64 @llvm.umin.i64(i64 %a4, i64 %b4)
+  %r5 = call i64 @llvm.umin.i64(i64 %a5, i64 %b5)
+  %r6 = call i64 @llvm.umin.i64(i64 %a6, i64 %b6)
+  %r7 = call i64 @llvm.umin.i64(i64 %a7, i64 %b7)
+  store i64 %r0, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 0), align 8
+  store i64 %r1, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 1), align 8
+  store i64 %r2, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 2), align 8
+  store i64 %r3, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 3), align 8
+  store i64 %r4, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 4), align 8
+  store i64 %r5, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 5), align 8
+  store i64 %r6, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 6), align 8
+  store i64 %r7, i64* getelementptr inbounds ([8 x i64], [8 x i64]* @c64, i32 0, i64 7), align 8
+  ret void
+}
+
+define void @umin_v16i32() {
+; SSE-LABEL: @umin_v16i32(
+; SSE-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP5:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @b32 to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP6:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP7:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP8:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    [[TMP9:%.*]] = call <4 x i32> @llvm.umin.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <4 x i32> @llvm.umin.v4i32(<4 x i32> [[TMP2]], <4 x i32> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <4 x i32> @llvm.umin.v4i32(<4 x i32> [[TMP3]], <4 x i32> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.umin.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP8]])
+; SSE-NEXT:    store <4 x i32> [[TMP9]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP10]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP11]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4
+; SSE-NEXT:    store <4 x i32> [[TMP12]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @umin_v16i32(
+; SLM-NEXT:    [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @a32 to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP3:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP4:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP5:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([16 x i32]* @b32 to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP6:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP7:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP8:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    [[TMP9:%.*]] = call <4 x i32> @llvm.umin.v4i32(<4 x i32> [[TMP1]], <4 x i32> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <4 x i32> @llvm.umin.v4i32(<4 x i32> [[TMP2]], <4 x i32> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <4 x i32> @llvm.umin.v4i32(<4 x i32> [[TMP3]], <4 x i32> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.umin.v4i32(<4 x i32> [[TMP4]], <4 x i32> [[TMP8]])
+; SLM-NEXT:    store <4 x i32> [[TMP9]], <4 x i32>* bitcast ([16 x i32]* @c32 to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP10]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4) to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP11]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <4 x i32>*), align 4
+; SLM-NEXT:    store <4 x i32> [[TMP12]], <4 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12) to <4 x i32>*), align 4
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @umin_v16i32(
+; AVX-NEXT:    [[TMP1:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([16 x i32]* @a32 to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP3:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([16 x i32]* @b32 to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP4:%.*]] = load <8 x i32>, <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    [[TMP5:%.*]] = call <8 x i32> @llvm.umin.v8i32(<8 x i32> [[TMP1]], <8 x i32> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <8 x i32> @llvm.umin.v8i32(<8 x i32> [[TMP2]], <8 x i32> [[TMP4]])
+; AVX-NEXT:    store <8 x i32> [[TMP5]], <8 x i32>* bitcast ([16 x i32]* @c32 to <8 x i32>*), align 4
+; AVX-NEXT:    store <8 x i32> [[TMP6]], <8 x i32>* bitcast (i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8) to <8 x i32>*), align 4
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @umin_v16i32(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <16 x i32>, <16 x i32>* bitcast ([16 x i32]* @a32 to <16 x i32>*), align 4
+; AVX512-NEXT:    [[TMP2:%.*]] = load <16 x i32>, <16 x i32>* bitcast ([16 x i32]* @b32 to <16 x i32>*), align 4
+; AVX512-NEXT:    [[TMP3:%.*]] = call <16 x i32> @llvm.umin.v16i32(<16 x i32> [[TMP1]], <16 x i32> [[TMP2]])
+; AVX512-NEXT:    store <16 x i32> [[TMP3]], <16 x i32>* bitcast ([16 x i32]* @c32 to <16 x i32>*), align 4
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 0 ), align 4
+  %a1  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 1 ), align 4
+  %a2  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 2 ), align 4
+  %a3  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 3 ), align 4
+  %a4  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 4 ), align 4
+  %a5  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 5 ), align 4
+  %a6  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 6 ), align 4
+  %a7  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 7 ), align 4
+  %a8  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 8 ), align 4
+  %a9  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 9 ), align 4
+  %a10 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 10), align 4
+  %a11 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 11), align 4
+  %a12 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 12), align 4
+  %a13 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 13), align 4
+  %a14 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 14), align 4
+  %a15 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @a32, i32 0, i64 15), align 4
+  %b0  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 0 ), align 4
+  %b1  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 1 ), align 4
+  %b2  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 2 ), align 4
+  %b3  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 3 ), align 4
+  %b4  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 4 ), align 4
+  %b5  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 5 ), align 4
+  %b6  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 6 ), align 4
+  %b7  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 7 ), align 4
+  %b8  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 8 ), align 4
+  %b9  = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 9 ), align 4
+  %b10 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 10), align 4
+  %b11 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 11), align 4
+  %b12 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 12), align 4
+  %b13 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 13), align 4
+  %b14 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 14), align 4
+  %b15 = load i32, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @b32, i32 0, i64 15), align 4
+  %r0  = call i32 @llvm.umin.i32(i32 %a0 , i32 %b0 )
+  %r1  = call i32 @llvm.umin.i32(i32 %a1 , i32 %b1 )
+  %r2  = call i32 @llvm.umin.i32(i32 %a2 , i32 %b2 )
+  %r3  = call i32 @llvm.umin.i32(i32 %a3 , i32 %b3 )
+  %r4  = call i32 @llvm.umin.i32(i32 %a4 , i32 %b4 )
+  %r5  = call i32 @llvm.umin.i32(i32 %a5 , i32 %b5 )
+  %r6  = call i32 @llvm.umin.i32(i32 %a6 , i32 %b6 )
+  %r7  = call i32 @llvm.umin.i32(i32 %a7 , i32 %b7 )
+  %r8  = call i32 @llvm.umin.i32(i32 %a8 , i32 %b8 )
+  %r9  = call i32 @llvm.umin.i32(i32 %a9 , i32 %b9 )
+  %r10 = call i32 @llvm.umin.i32(i32 %a10, i32 %b10)
+  %r11 = call i32 @llvm.umin.i32(i32 %a11, i32 %b11)
+  %r12 = call i32 @llvm.umin.i32(i32 %a12, i32 %b12)
+  %r13 = call i32 @llvm.umin.i32(i32 %a13, i32 %b13)
+  %r14 = call i32 @llvm.umin.i32(i32 %a14, i32 %b14)
+  %r15 = call i32 @llvm.umin.i32(i32 %a15, i32 %b15)
+  store i32 %r0 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 0 ), align 4
+  store i32 %r1 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 1 ), align 4
+  store i32 %r2 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 2 ), align 4
+  store i32 %r3 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 3 ), align 4
+  store i32 %r4 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 4 ), align 4
+  store i32 %r5 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 5 ), align 4
+  store i32 %r6 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 6 ), align 4
+  store i32 %r7 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 7 ), align 4
+  store i32 %r8 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 8 ), align 4
+  store i32 %r9 , i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 9 ), align 4
+  store i32 %r10, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 10), align 4
+  store i32 %r11, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 11), align 4
+  store i32 %r12, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 12), align 4
+  store i32 %r13, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 13), align 4
+  store i32 %r14, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 14), align 4
+  store i32 %r15, i32* getelementptr inbounds ([16 x i32], [16 x i32]* @c32, i32 0, i64 15), align 4
+  ret void
+}
+
+define void @umin_v32i16() {
+; SSE-LABEL: @umin_v32i16(
+; SSE-NEXT:    [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @a16 to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP3:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP5:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @b16 to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP6:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP7:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP8:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    [[TMP9:%.*]] = call <8 x i16> @llvm.umin.v8i16(<8 x i16> [[TMP1]], <8 x i16> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <8 x i16> @llvm.umin.v8i16(<8 x i16> [[TMP2]], <8 x i16> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <8 x i16> @llvm.umin.v8i16(<8 x i16> [[TMP3]], <8 x i16> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <8 x i16> @llvm.umin.v8i16(<8 x i16> [[TMP4]], <8 x i16> [[TMP8]])
+; SSE-NEXT:    store <8 x i16> [[TMP9]], <8 x i16>* bitcast ([32 x i16]* @c16 to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP10]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP11]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <8 x i16>*), align 2
+; SSE-NEXT:    store <8 x i16> [[TMP12]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24) to <8 x i16>*), align 2
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @umin_v32i16(
+; SLM-NEXT:    [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @a16 to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP3:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP5:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([32 x i16]* @b16 to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP6:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP7:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP8:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    [[TMP9:%.*]] = call <8 x i16> @llvm.umin.v8i16(<8 x i16> [[TMP1]], <8 x i16> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <8 x i16> @llvm.umin.v8i16(<8 x i16> [[TMP2]], <8 x i16> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <8 x i16> @llvm.umin.v8i16(<8 x i16> [[TMP3]], <8 x i16> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <8 x i16> @llvm.umin.v8i16(<8 x i16> [[TMP4]], <8 x i16> [[TMP8]])
+; SLM-NEXT:    store <8 x i16> [[TMP9]], <8 x i16>* bitcast ([32 x i16]* @c16 to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP10]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8) to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP11]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <8 x i16>*), align 2
+; SLM-NEXT:    store <8 x i16> [[TMP12]], <8 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24) to <8 x i16>*), align 2
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @umin_v32i16(
+; AVX-NEXT:    [[TMP1:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([32 x i16]* @a16 to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP2:%.*]] = load <16 x i16>, <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP3:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([32 x i16]* @b16 to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP4:%.*]] = load <16 x i16>, <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    [[TMP5:%.*]] = call <16 x i16> @llvm.umin.v16i16(<16 x i16> [[TMP1]], <16 x i16> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <16 x i16> @llvm.umin.v16i16(<16 x i16> [[TMP2]], <16 x i16> [[TMP4]])
+; AVX-NEXT:    store <16 x i16> [[TMP5]], <16 x i16>* bitcast ([32 x i16]* @c16 to <16 x i16>*), align 2
+; AVX-NEXT:    store <16 x i16> [[TMP6]], <16 x i16>* bitcast (i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16) to <16 x i16>*), align 2
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @umin_v32i16(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <32 x i16>, <32 x i16>* bitcast ([32 x i16]* @a16 to <32 x i16>*), align 2
+; AVX512-NEXT:    [[TMP2:%.*]] = load <32 x i16>, <32 x i16>* bitcast ([32 x i16]* @b16 to <32 x i16>*), align 2
+; AVX512-NEXT:    [[TMP3:%.*]] = call <32 x i16> @llvm.umin.v32i16(<32 x i16> [[TMP1]], <32 x i16> [[TMP2]])
+; AVX512-NEXT:    store <32 x i16> [[TMP3]], <32 x i16>* bitcast ([32 x i16]* @c16 to <32 x i16>*), align 2
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 0 ), align 2
+  %a1  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 1 ), align 2
+  %a2  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 2 ), align 2
+  %a3  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 3 ), align 2
+  %a4  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 4 ), align 2
+  %a5  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 5 ), align 2
+  %a6  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 6 ), align 2
+  %a7  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 7 ), align 2
+  %a8  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 8 ), align 2
+  %a9  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 9 ), align 2
+  %a10 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 10), align 2
+  %a11 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 11), align 2
+  %a12 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 12), align 2
+  %a13 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 13), align 2
+  %a14 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 14), align 2
+  %a15 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 15), align 2
+  %a16 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 16), align 2
+  %a17 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 17), align 2
+  %a18 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 18), align 2
+  %a19 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 19), align 2
+  %a20 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 20), align 2
+  %a21 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 21), align 2
+  %a22 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 22), align 2
+  %a23 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 23), align 2
+  %a24 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 24), align 2
+  %a25 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 25), align 2
+  %a26 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 26), align 2
+  %a27 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 27), align 2
+  %a28 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 28), align 2
+  %a29 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 29), align 2
+  %a30 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 30), align 2
+  %a31 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @a16, i32 0, i64 31), align 2
+  %b0  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 0 ), align 2
+  %b1  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 1 ), align 2
+  %b2  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 2 ), align 2
+  %b3  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 3 ), align 2
+  %b4  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 4 ), align 2
+  %b5  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 5 ), align 2
+  %b6  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 6 ), align 2
+  %b7  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 7 ), align 2
+  %b8  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 8 ), align 2
+  %b9  = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 9 ), align 2
+  %b10 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 10), align 2
+  %b11 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 11), align 2
+  %b12 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 12), align 2
+  %b13 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 13), align 2
+  %b14 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 14), align 2
+  %b15 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 15), align 2
+  %b16 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 16), align 2
+  %b17 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 17), align 2
+  %b18 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 18), align 2
+  %b19 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 19), align 2
+  %b20 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 20), align 2
+  %b21 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 21), align 2
+  %b22 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 22), align 2
+  %b23 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 23), align 2
+  %b24 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 24), align 2
+  %b25 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 25), align 2
+  %b26 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 26), align 2
+  %b27 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 27), align 2
+  %b28 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 28), align 2
+  %b29 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 29), align 2
+  %b30 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 30), align 2
+  %b31 = load i16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @b16, i32 0, i64 31), align 2
+  %r0  = call i16 @llvm.umin.i16(i16 %a0 , i16 %b0 )
+  %r1  = call i16 @llvm.umin.i16(i16 %a1 , i16 %b1 )
+  %r2  = call i16 @llvm.umin.i16(i16 %a2 , i16 %b2 )
+  %r3  = call i16 @llvm.umin.i16(i16 %a3 , i16 %b3 )
+  %r4  = call i16 @llvm.umin.i16(i16 %a4 , i16 %b4 )
+  %r5  = call i16 @llvm.umin.i16(i16 %a5 , i16 %b5 )
+  %r6  = call i16 @llvm.umin.i16(i16 %a6 , i16 %b6 )
+  %r7  = call i16 @llvm.umin.i16(i16 %a7 , i16 %b7 )
+  %r8  = call i16 @llvm.umin.i16(i16 %a8 , i16 %b8 )
+  %r9  = call i16 @llvm.umin.i16(i16 %a9 , i16 %b9 )
+  %r10 = call i16 @llvm.umin.i16(i16 %a10, i16 %b10)
+  %r11 = call i16 @llvm.umin.i16(i16 %a11, i16 %b11)
+  %r12 = call i16 @llvm.umin.i16(i16 %a12, i16 %b12)
+  %r13 = call i16 @llvm.umin.i16(i16 %a13, i16 %b13)
+  %r14 = call i16 @llvm.umin.i16(i16 %a14, i16 %b14)
+  %r15 = call i16 @llvm.umin.i16(i16 %a15, i16 %b15)
+  %r16 = call i16 @llvm.umin.i16(i16 %a16, i16 %b16)
+  %r17 = call i16 @llvm.umin.i16(i16 %a17, i16 %b17)
+  %r18 = call i16 @llvm.umin.i16(i16 %a18, i16 %b18)
+  %r19 = call i16 @llvm.umin.i16(i16 %a19, i16 %b19)
+  %r20 = call i16 @llvm.umin.i16(i16 %a20, i16 %b20)
+  %r21 = call i16 @llvm.umin.i16(i16 %a21, i16 %b21)
+  %r22 = call i16 @llvm.umin.i16(i16 %a22, i16 %b22)
+  %r23 = call i16 @llvm.umin.i16(i16 %a23, i16 %b23)
+  %r24 = call i16 @llvm.umin.i16(i16 %a24, i16 %b24)
+  %r25 = call i16 @llvm.umin.i16(i16 %a25, i16 %b25)
+  %r26 = call i16 @llvm.umin.i16(i16 %a26, i16 %b26)
+  %r27 = call i16 @llvm.umin.i16(i16 %a27, i16 %b27)
+  %r28 = call i16 @llvm.umin.i16(i16 %a28, i16 %b28)
+  %r29 = call i16 @llvm.umin.i16(i16 %a29, i16 %b29)
+  %r30 = call i16 @llvm.umin.i16(i16 %a30, i16 %b30)
+  %r31 = call i16 @llvm.umin.i16(i16 %a31, i16 %b31)
+  store i16 %r0 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 0 ), align 2
+  store i16 %r1 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 1 ), align 2
+  store i16 %r2 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 2 ), align 2
+  store i16 %r3 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 3 ), align 2
+  store i16 %r4 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 4 ), align 2
+  store i16 %r5 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 5 ), align 2
+  store i16 %r6 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 6 ), align 2
+  store i16 %r7 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 7 ), align 2
+  store i16 %r8 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 8 ), align 2
+  store i16 %r9 , i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 9 ), align 2
+  store i16 %r10, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 10), align 2
+  store i16 %r11, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 11), align 2
+  store i16 %r12, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 12), align 2
+  store i16 %r13, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 13), align 2
+  store i16 %r14, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 14), align 2
+  store i16 %r15, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 15), align 2
+  store i16 %r16, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 16), align 2
+  store i16 %r17, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 17), align 2
+  store i16 %r18, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 18), align 2
+  store i16 %r19, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 19), align 2
+  store i16 %r20, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 20), align 2
+  store i16 %r21, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 21), align 2
+  store i16 %r22, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 22), align 2
+  store i16 %r23, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 23), align 2
+  store i16 %r24, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 24), align 2
+  store i16 %r25, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 25), align 2
+  store i16 %r26, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 26), align 2
+  store i16 %r27, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 27), align 2
+  store i16 %r28, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 28), align 2
+  store i16 %r29, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 29), align 2
+  store i16 %r30, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 30), align 2
+  store i16 %r31, i16* getelementptr inbounds ([32 x i16], [32 x i16]* @c16, i32 0, i64 31), align 2
+  ret void
+}
+
+define void @umin_v64i8() {
+; SSE-LABEL: @umin_v64i8(
+; SSE-NEXT:    [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @a8 to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP5:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @b8 to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP6:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP7:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP8:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    [[TMP9:%.*]] = call <16 x i8> @llvm.umin.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP5]])
+; SSE-NEXT:    [[TMP10:%.*]] = call <16 x i8> @llvm.umin.v16i8(<16 x i8> [[TMP2]], <16 x i8> [[TMP6]])
+; SSE-NEXT:    [[TMP11:%.*]] = call <16 x i8> @llvm.umin.v16i8(<16 x i8> [[TMP3]], <16 x i8> [[TMP7]])
+; SSE-NEXT:    [[TMP12:%.*]] = call <16 x i8> @llvm.umin.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP8]])
+; SSE-NEXT:    store <16 x i8> [[TMP9]], <16 x i8>* bitcast ([64 x i8]* @c8 to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP10]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16) to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP11]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <16 x i8>*), align 1
+; SSE-NEXT:    store <16 x i8> [[TMP12]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48) to <16 x i8>*), align 1
+; SSE-NEXT:    ret void
+;
+; SLM-LABEL: @umin_v64i8(
+; SLM-NEXT:    [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @a8 to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP5:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([64 x i8]* @b8 to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP6:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP7:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP8:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    [[TMP9:%.*]] = call <16 x i8> @llvm.umin.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP5]])
+; SLM-NEXT:    [[TMP10:%.*]] = call <16 x i8> @llvm.umin.v16i8(<16 x i8> [[TMP2]], <16 x i8> [[TMP6]])
+; SLM-NEXT:    [[TMP11:%.*]] = call <16 x i8> @llvm.umin.v16i8(<16 x i8> [[TMP3]], <16 x i8> [[TMP7]])
+; SLM-NEXT:    [[TMP12:%.*]] = call <16 x i8> @llvm.umin.v16i8(<16 x i8> [[TMP4]], <16 x i8> [[TMP8]])
+; SLM-NEXT:    store <16 x i8> [[TMP9]], <16 x i8>* bitcast ([64 x i8]* @c8 to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP10]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16) to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP11]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <16 x i8>*), align 1
+; SLM-NEXT:    store <16 x i8> [[TMP12]], <16 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48) to <16 x i8>*), align 1
+; SLM-NEXT:    ret void
+;
+; AVX-LABEL: @umin_v64i8(
+; AVX-NEXT:    [[TMP1:%.*]] = load <32 x i8>, <32 x i8>* bitcast ([64 x i8]* @a8 to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP2:%.*]] = load <32 x i8>, <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP3:%.*]] = load <32 x i8>, <32 x i8>* bitcast ([64 x i8]* @b8 to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP4:%.*]] = load <32 x i8>, <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    [[TMP5:%.*]] = call <32 x i8> @llvm.umin.v32i8(<32 x i8> [[TMP1]], <32 x i8> [[TMP3]])
+; AVX-NEXT:    [[TMP6:%.*]] = call <32 x i8> @llvm.umin.v32i8(<32 x i8> [[TMP2]], <32 x i8> [[TMP4]])
+; AVX-NEXT:    store <32 x i8> [[TMP5]], <32 x i8>* bitcast ([64 x i8]* @c8 to <32 x i8>*), align 1
+; AVX-NEXT:    store <32 x i8> [[TMP6]], <32 x i8>* bitcast (i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32) to <32 x i8>*), align 1
+; AVX-NEXT:    ret void
+;
+; AVX512-LABEL: @umin_v64i8(
+; AVX512-NEXT:    [[TMP1:%.*]] = load <64 x i8>, <64 x i8>* bitcast ([64 x i8]* @a8 to <64 x i8>*), align 1
+; AVX512-NEXT:    [[TMP2:%.*]] = load <64 x i8>, <64 x i8>* bitcast ([64 x i8]* @b8 to <64 x i8>*), align 1
+; AVX512-NEXT:    [[TMP3:%.*]] = call <64 x i8> @llvm.umin.v64i8(<64 x i8> [[TMP1]], <64 x i8> [[TMP2]])
+; AVX512-NEXT:    store <64 x i8> [[TMP3]], <64 x i8>* bitcast ([64 x i8]* @c8 to <64 x i8>*), align 1
+; AVX512-NEXT:    ret void
+;
+  %a0  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 0 ), align 1
+  %a1  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 1 ), align 1
+  %a2  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 2 ), align 1
+  %a3  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 3 ), align 1
+  %a4  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 4 ), align 1
+  %a5  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 5 ), align 1
+  %a6  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 6 ), align 1
+  %a7  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 7 ), align 1
+  %a8  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 8 ), align 1
+  %a9  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 9 ), align 1
+  %a10 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 10), align 1
+  %a11 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 11), align 1
+  %a12 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 12), align 1
+  %a13 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 13), align 1
+  %a14 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 14), align 1
+  %a15 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 15), align 1
+  %a16 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 16), align 1
+  %a17 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 17), align 1
+  %a18 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 18), align 1
+  %a19 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 19), align 1
+  %a20 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 20), align 1
+  %a21 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 21), align 1
+  %a22 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 22), align 1
+  %a23 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 23), align 1
+  %a24 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 24), align 1
+  %a25 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 25), align 1
+  %a26 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 26), align 1
+  %a27 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 27), align 1
+  %a28 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 28), align 1
+  %a29 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 29), align 1
+  %a30 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 30), align 1
+  %a31 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 31), align 1
+  %a32 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 32), align 1
+  %a33 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 33), align 1
+  %a34 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 34), align 1
+  %a35 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 35), align 1
+  %a36 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 36), align 1
+  %a37 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 37), align 1
+  %a38 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 38), align 1
+  %a39 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 39), align 1
+  %a40 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 40), align 1
+  %a41 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 41), align 1
+  %a42 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 42), align 1
+  %a43 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 43), align 1
+  %a44 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 44), align 1
+  %a45 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 45), align 1
+  %a46 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 46), align 1
+  %a47 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 47), align 1
+  %a48 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 48), align 1
+  %a49 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 49), align 1
+  %a50 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 50), align 1
+  %a51 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 51), align 1
+  %a52 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 52), align 1
+  %a53 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 53), align 1
+  %a54 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 54), align 1
+  %a55 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 55), align 1
+  %a56 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 56), align 1
+  %a57 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 57), align 1
+  %a58 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 58), align 1
+  %a59 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 59), align 1
+  %a60 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 60), align 1
+  %a61 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 61), align 1
+  %a62 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 62), align 1
+  %a63 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @a8, i32 0, i64 63), align 1
+  %b0  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 0 ), align 1
+  %b1  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 1 ), align 1
+  %b2  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 2 ), align 1
+  %b3  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 3 ), align 1
+  %b4  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 4 ), align 1
+  %b5  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 5 ), align 1
+  %b6  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 6 ), align 1
+  %b7  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 7 ), align 1
+  %b8  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 8 ), align 1
+  %b9  = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 9 ), align 1
+  %b10 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 10), align 1
+  %b11 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 11), align 1
+  %b12 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 12), align 1
+  %b13 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 13), align 1
+  %b14 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 14), align 1
+  %b15 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 15), align 1
+  %b16 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 16), align 1
+  %b17 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 17), align 1
+  %b18 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 18), align 1
+  %b19 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 19), align 1
+  %b20 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 20), align 1
+  %b21 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 21), align 1
+  %b22 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 22), align 1
+  %b23 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 23), align 1
+  %b24 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 24), align 1
+  %b25 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 25), align 1
+  %b26 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 26), align 1
+  %b27 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 27), align 1
+  %b28 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 28), align 1
+  %b29 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 29), align 1
+  %b30 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 30), align 1
+  %b31 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 31), align 1
+  %b32 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 32), align 1
+  %b33 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 33), align 1
+  %b34 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 34), align 1
+  %b35 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 35), align 1
+  %b36 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 36), align 1
+  %b37 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 37), align 1
+  %b38 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 38), align 1
+  %b39 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 39), align 1
+  %b40 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 40), align 1
+  %b41 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 41), align 1
+  %b42 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 42), align 1
+  %b43 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 43), align 1
+  %b44 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 44), align 1
+  %b45 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 45), align 1
+  %b46 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 46), align 1
+  %b47 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 47), align 1
+  %b48 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 48), align 1
+  %b49 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 49), align 1
+  %b50 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 50), align 1
+  %b51 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 51), align 1
+  %b52 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 52), align 1
+  %b53 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 53), align 1
+  %b54 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 54), align 1
+  %b55 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 55), align 1
+  %b56 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 56), align 1
+  %b57 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 57), align 1
+  %b58 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 58), align 1
+  %b59 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 59), align 1
+  %b60 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 60), align 1
+  %b61 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 61), align 1
+  %b62 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 62), align 1
+  %b63 = load i8, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @b8, i32 0, i64 63), align 1
+  %r0  = call i8 @llvm.umin.i8(i8 %a0 , i8 %b0 )
+  %r1  = call i8 @llvm.umin.i8(i8 %a1 , i8 %b1 )
+  %r2  = call i8 @llvm.umin.i8(i8 %a2 , i8 %b2 )
+  %r3  = call i8 @llvm.umin.i8(i8 %a3 , i8 %b3 )
+  %r4  = call i8 @llvm.umin.i8(i8 %a4 , i8 %b4 )
+  %r5  = call i8 @llvm.umin.i8(i8 %a5 , i8 %b5 )
+  %r6  = call i8 @llvm.umin.i8(i8 %a6 , i8 %b6 )
+  %r7  = call i8 @llvm.umin.i8(i8 %a7 , i8 %b7 )
+  %r8  = call i8 @llvm.umin.i8(i8 %a8 , i8 %b8 )
+  %r9  = call i8 @llvm.umin.i8(i8 %a9 , i8 %b9 )
+  %r10 = call i8 @llvm.umin.i8(i8 %a10, i8 %b10)
+  %r11 = call i8 @llvm.umin.i8(i8 %a11, i8 %b11)
+  %r12 = call i8 @llvm.umin.i8(i8 %a12, i8 %b12)
+  %r13 = call i8 @llvm.umin.i8(i8 %a13, i8 %b13)
+  %r14 = call i8 @llvm.umin.i8(i8 %a14, i8 %b14)
+  %r15 = call i8 @llvm.umin.i8(i8 %a15, i8 %b15)
+  %r16 = call i8 @llvm.umin.i8(i8 %a16, i8 %b16)
+  %r17 = call i8 @llvm.umin.i8(i8 %a17, i8 %b17)
+  %r18 = call i8 @llvm.umin.i8(i8 %a18, i8 %b18)
+  %r19 = call i8 @llvm.umin.i8(i8 %a19, i8 %b19)
+  %r20 = call i8 @llvm.umin.i8(i8 %a20, i8 %b20)
+  %r21 = call i8 @llvm.umin.i8(i8 %a21, i8 %b21)
+  %r22 = call i8 @llvm.umin.i8(i8 %a22, i8 %b22)
+  %r23 = call i8 @llvm.umin.i8(i8 %a23, i8 %b23)
+  %r24 = call i8 @llvm.umin.i8(i8 %a24, i8 %b24)
+  %r25 = call i8 @llvm.umin.i8(i8 %a25, i8 %b25)
+  %r26 = call i8 @llvm.umin.i8(i8 %a26, i8 %b26)
+  %r27 = call i8 @llvm.umin.i8(i8 %a27, i8 %b27)
+  %r28 = call i8 @llvm.umin.i8(i8 %a28, i8 %b28)
+  %r29 = call i8 @llvm.umin.i8(i8 %a29, i8 %b29)
+  %r30 = call i8 @llvm.umin.i8(i8 %a30, i8 %b30)
+  %r31 = call i8 @llvm.umin.i8(i8 %a31, i8 %b31)
+  %r32 = call i8 @llvm.umin.i8(i8 %a32, i8 %b32)
+  %r33 = call i8 @llvm.umin.i8(i8 %a33, i8 %b33)
+  %r34 = call i8 @llvm.umin.i8(i8 %a34, i8 %b34)
+  %r35 = call i8 @llvm.umin.i8(i8 %a35, i8 %b35)
+  %r36 = call i8 @llvm.umin.i8(i8 %a36, i8 %b36)
+  %r37 = call i8 @llvm.umin.i8(i8 %a37, i8 %b37)
+  %r38 = call i8 @llvm.umin.i8(i8 %a38, i8 %b38)
+  %r39 = call i8 @llvm.umin.i8(i8 %a39, i8 %b39)
+  %r40 = call i8 @llvm.umin.i8(i8 %a40, i8 %b40)
+  %r41 = call i8 @llvm.umin.i8(i8 %a41, i8 %b41)
+  %r42 = call i8 @llvm.umin.i8(i8 %a42, i8 %b42)
+  %r43 = call i8 @llvm.umin.i8(i8 %a43, i8 %b43)
+  %r44 = call i8 @llvm.umin.i8(i8 %a44, i8 %b44)
+  %r45 = call i8 @llvm.umin.i8(i8 %a45, i8 %b45)
+  %r46 = call i8 @llvm.umin.i8(i8 %a46, i8 %b46)
+  %r47 = call i8 @llvm.umin.i8(i8 %a47, i8 %b47)
+  %r48 = call i8 @llvm.umin.i8(i8 %a48, i8 %b48)
+  %r49 = call i8 @llvm.umin.i8(i8 %a49, i8 %b49)
+  %r50 = call i8 @llvm.umin.i8(i8 %a50, i8 %b50)
+  %r51 = call i8 @llvm.umin.i8(i8 %a51, i8 %b51)
+  %r52 = call i8 @llvm.umin.i8(i8 %a52, i8 %b52)
+  %r53 = call i8 @llvm.umin.i8(i8 %a53, i8 %b53)
+  %r54 = call i8 @llvm.umin.i8(i8 %a54, i8 %b54)
+  %r55 = call i8 @llvm.umin.i8(i8 %a55, i8 %b55)
+  %r56 = call i8 @llvm.umin.i8(i8 %a56, i8 %b56)
+  %r57 = call i8 @llvm.umin.i8(i8 %a57, i8 %b57)
+  %r58 = call i8 @llvm.umin.i8(i8 %a58, i8 %b58)
+  %r59 = call i8 @llvm.umin.i8(i8 %a59, i8 %b59)
+  %r60 = call i8 @llvm.umin.i8(i8 %a60, i8 %b60)
+  %r61 = call i8 @llvm.umin.i8(i8 %a61, i8 %b61)
+  %r62 = call i8 @llvm.umin.i8(i8 %a62, i8 %b62)
+  %r63 = call i8 @llvm.umin.i8(i8 %a63, i8 %b63)
+  store i8 %r0 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 0 ), align 1
+  store i8 %r1 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 1 ), align 1
+  store i8 %r2 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 2 ), align 1
+  store i8 %r3 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 3 ), align 1
+  store i8 %r4 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 4 ), align 1
+  store i8 %r5 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 5 ), align 1
+  store i8 %r6 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 6 ), align 1
+  store i8 %r7 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 7 ), align 1
+  store i8 %r8 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 8 ), align 1
+  store i8 %r9 , i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 9 ), align 1
+  store i8 %r10, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 10), align 1
+  store i8 %r11, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 11), align 1
+  store i8 %r12, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 12), align 1
+  store i8 %r13, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 13), align 1
+  store i8 %r14, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 14), align 1
+  store i8 %r15, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 15), align 1
+  store i8 %r16, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 16), align 1
+  store i8 %r17, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 17), align 1
+  store i8 %r18, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 18), align 1
+  store i8 %r19, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 19), align 1
+  store i8 %r20, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 20), align 1
+  store i8 %r21, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 21), align 1
+  store i8 %r22, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 22), align 1
+  store i8 %r23, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 23), align 1
+  store i8 %r24, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 24), align 1
+  store i8 %r25, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 25), align 1
+  store i8 %r26, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 26), align 1
+  store i8 %r27, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 27), align 1
+  store i8 %r28, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 28), align 1
+  store i8 %r29, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 29), align 1
+  store i8 %r30, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 30), align 1
+  store i8 %r31, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 31), align 1
+  store i8 %r32, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 32), align 1
+  store i8 %r33, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 33), align 1
+  store i8 %r34, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 34), align 1
+  store i8 %r35, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 35), align 1
+  store i8 %r36, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 36), align 1
+  store i8 %r37, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 37), align 1
+  store i8 %r38, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 38), align 1
+  store i8 %r39, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 39), align 1
+  store i8 %r40, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 40), align 1
+  store i8 %r41, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 41), align 1
+  store i8 %r42, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 42), align 1
+  store i8 %r43, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 43), align 1
+  store i8 %r44, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 44), align 1
+  store i8 %r45, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 45), align 1
+  store i8 %r46, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 46), align 1
+  store i8 %r47, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 47), align 1
+  store i8 %r48, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 48), align 1
+  store i8 %r49, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 49), align 1
+  store i8 %r50, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 50), align 1
+  store i8 %r51, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 51), align 1
+  store i8 %r52, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 52), align 1
+  store i8 %r53, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 53), align 1
+  store i8 %r54, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 54), align 1
+  store i8 %r55, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 55), align 1
+  store i8 %r56, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 56), align 1
+  store i8 %r57, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 57), align 1
+  store i8 %r58, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 58), align 1
+  store i8 %r59, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 59), align 1
+  store i8 %r60, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 60), align 1
+  store i8 %r61, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 61), align 1
+  store i8 %r62, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 62), align 1
+  store i8 %r63, i8* getelementptr inbounds ([64 x i8], [64 x i8]* @c8, i32 0, i64 63), align 1
+  ret void
+}