RPCSX/llvm
1
0
Fork 0
mirror of https://github.com/RPCSX/llvm.git synced 2025-01-11 15:07:58 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

[AVX-512] Remove and autoupgrade masked dword/qword variable shift intrinsics to the new unmasked versions and selects.

Browse Source 
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@286786 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Craig Topper 2016-11-14 01:53:22 +00:00
parent 00eb18da3b
commit d03c7ac104
7 changed files with 291 additions and 326 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

38
include/llvm/IR/IntrinsicsX86.td
View File

@ -2358,38 +2358,6 @@ let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.".
Intrinsic<[llvm_v8i64_ty], [llvm_v8i64_ty, llvm_v8i64_ty],
[IntrNoMem]>;
def int_x86_avx512_mask_psllv_d : GCCBuiltin<"__builtin_ia32_psllv16si_mask">,
Intrinsic<[llvm_v16i32_ty], [llvm_v16i32_ty,
llvm_v16i32_ty, llvm_v16i32_ty, llvm_i16_ty],
[IntrNoMem]>;
def int_x86_avx512_mask_psllv_q : GCCBuiltin<"__builtin_ia32_psllv8di_mask">,
Intrinsic<[llvm_v8i64_ty], [llvm_v8i64_ty,
llvm_v8i64_ty, llvm_v8i64_ty, llvm_i8_ty],
[IntrNoMem]>;
def int_x86_avx512_mask_psrav_d : GCCBuiltin<"__builtin_ia32_psrav16si_mask">,
Intrinsic<[llvm_v16i32_ty], [llvm_v16i32_ty,
llvm_v16i32_ty, llvm_v16i32_ty, llvm_i16_ty],
[IntrNoMem]>;
def int_x86_avx512_mask_psrav_q : GCCBuiltin<"__builtin_ia32_psrav8di_mask">,
Intrinsic<[llvm_v8i64_ty], [llvm_v8i64_ty,
llvm_v8i64_ty, llvm_v8i64_ty, llvm_i8_ty],
[IntrNoMem]>;
def int_x86_avx512_mask_psrlv_d : GCCBuiltin<"__builtin_ia32_psrlv16si_mask">,
Intrinsic<[llvm_v16i32_ty], [llvm_v16i32_ty,
llvm_v16i32_ty, llvm_v16i32_ty, llvm_i16_ty],
[IntrNoMem]>;
def int_x86_avx512_mask_psrlv_q : GCCBuiltin<"__builtin_ia32_psrlv8di_mask">,
Intrinsic<[llvm_v8i64_ty], [llvm_v8i64_ty,
llvm_v8i64_ty, llvm_v8i64_ty, llvm_i8_ty],
[IntrNoMem]>;
def int_x86_avx512_mask_psll_di_512 : GCCBuiltin<"__builtin_ia32_pslldi512_mask">,
Intrinsic<[llvm_v16i32_ty], [llvm_v16i32_ty,
llvm_i32_ty, llvm_v16i32_ty, llvm_i16_ty], [IntrNoMem]>;
def int_x86_avx512_mask_psll_qi_512 : GCCBuiltin<"__builtin_ia32_psllqi512_mask">,
Intrinsic<[llvm_v8i64_ty], [llvm_v8i64_ty,
llvm_i32_ty, llvm_v8i64_ty, llvm_i8_ty], [IntrNoMem]>;
def int_x86_avx512_mask_psrav16_hi : GCCBuiltin<"__builtin_ia32_psrav16hi_mask">,
Intrinsic<[llvm_v16i16_ty], [llvm_v16i16_ty,
llvm_v16i16_ty, llvm_v16i16_ty, llvm_i16_ty], [IntrNoMem]>;
@ -2399,12 +2367,6 @@ let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.".
def int_x86_avx512_mask_psrav8_hi : GCCBuiltin<"__builtin_ia32_psrav8hi_mask">,
Intrinsic<[llvm_v8i16_ty], [llvm_v8i16_ty,
llvm_v8i16_ty, llvm_v8i16_ty, llvm_i8_ty], [IntrNoMem]>;
def int_x86_avx512_mask_psrav_q_128 : GCCBuiltin<"__builtin_ia32_psravq128_mask">,
Intrinsic<[llvm_v2i64_ty], [llvm_v2i64_ty,
llvm_v2i64_ty, llvm_v2i64_ty, llvm_i8_ty], [IntrNoMem]>;
def int_x86_avx512_mask_psrav_q_256 : GCCBuiltin<"__builtin_ia32_psravq256_mask">,
Intrinsic<[llvm_v4i64_ty], [llvm_v4i64_ty,
llvm_v4i64_ty, llvm_v4i64_ty, llvm_i8_ty], [IntrNoMem]>;
def int_x86_avx512_mask_psrlv16_hi : GCCBuiltin<"__builtin_ia32_psrlv16hi_mask">,
Intrinsic<[llvm_v16i16_ty], [llvm_v16i16_ty,

59
lib/IR/AutoUpgrade.cpp
View File

@ -320,6 +320,9 @@ static bool UpgradeIntrinsicFunction1(Function *F, Function *&NewFn) {
Name == "avx512.mask.psrlv4.di" ||
Name == "avx512.mask.psrlv4.si" ||
Name == "avx512.mask.psrlv8.si" ||
Name.startswith("avx512.mask.psllv.") ||
Name.startswith("avx512.mask.psrav.") ||
Name.startswith("avx512.mask.psrlv.") ||
Name.startswith("sse41.pmovsx") ||
Name.startswith("sse41.pmovzx") ||
Name.startswith("avx2.pmovsx") ||
@ -1364,7 +1367,7 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
Name[19];
Intrinsic::ID IID;
if (IsVariable) {
if (IsVariable && Name[17] != '.') {
if (Size == 'd' && Name[17] == '2') // avx512.mask.psllv2.di
IID = Intrinsic::x86_avx2_psllv_q;
else if (Size == 'd' && Name[17] == '4') // avx512.mask.psllv4.di
@ -1400,12 +1403,14 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
else
llvm_unreachable("Unexpected size");
} else {
if (Size == 'd') // psll.di.512, pslli.d, psll.d
IID = IsImmediate ? Intrinsic::x86_avx512_pslli_d_512
: Intrinsic::x86_avx512_psll_d_512;
else if (Size == 'q') // psll.qi.512, pslli.q, psll.q
IID = IsImmediate ? Intrinsic::x86_avx512_pslli_q_512
: Intrinsic::x86_avx512_psll_q_512;
if (Size == 'd') // psll.di.512, pslli.d, psll.d, psllv.d.512
IID = IsImmediate ? Intrinsic::x86_avx512_pslli_d_512 :
IsVariable ? Intrinsic::x86_avx512_psllv_d_512 :
Intrinsic::x86_avx512_psll_d_512;
else if (Size == 'q') // psll.qi.512, pslli.q, psll.q, psllv.q.512
IID = IsImmediate ? Intrinsic::x86_avx512_pslli_q_512 :
IsVariable ? Intrinsic::x86_avx512_psllv_q_512 :
Intrinsic::x86_avx512_psll_q_512;
else if (Size == 'w') // psll.wi.512, pslli.w, psll.w
IID = IsImmediate ? Intrinsic::x86_avx512_pslli_w_512
: Intrinsic::x86_avx512_psll_w_512;
@ -1423,7 +1428,7 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
Name[19];
Intrinsic::ID IID;
if (IsVariable) {
if (IsVariable && Name[17] != '.') {
if (Size == 'd' && Name[17] == '2') // avx512.mask.psrlv2.di
IID = Intrinsic::x86_avx2_psrlv_q;
else if (Size == 'd' && Name[17] == '4') // avx512.mask.psrlv4.di
@ -1459,12 +1464,14 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
else
llvm_unreachable("Unexpected size");
} else {
if (Size == 'd') // psrl.di.512, psrli.d, psrl.d
IID = IsImmediate ? Intrinsic::x86_avx512_psrli_d_512
: Intrinsic::x86_avx512_psrl_d_512;
else if (Size == 'q') // psrl.qi.512, psrli.q, psrl.q
IID = IsImmediate ? Intrinsic::x86_avx512_psrli_q_512
: Intrinsic::x86_avx512_psrl_q_512;
if (Size == 'd') // psrl.di.512, psrli.d, psrl.d, psrl.d.512
IID = IsImmediate ? Intrinsic::x86_avx512_psrli_d_512 :
IsVariable ? Intrinsic::x86_avx512_psrlv_d_512 :
Intrinsic::x86_avx512_psrl_d_512;
else if (Size == 'q') // psrl.qi.512, psrli.q, psrl.q, psrl.q.512
IID = IsImmediate ? Intrinsic::x86_avx512_psrli_q_512 :
IsVariable ? Intrinsic::x86_avx512_psrlv_q_512 :
Intrinsic::x86_avx512_psrl_q_512;
else if (Size == 'w') // psrl.wi.512, psrli.w, psrl.w)
IID = IsImmediate ? Intrinsic::x86_avx512_psrli_w_512
: Intrinsic::x86_avx512_psrl_w_512;
@ -1482,7 +1489,7 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
Name[19];
Intrinsic::ID IID;
if (IsVariable) {
if (IsVariable && Name[17] != '.') {
if (Size == 's' && Name[17] == '4') // avx512.mask.psrav4.si
IID = Intrinsic::x86_avx2_psrav_d;
else if (Size == 's' && Name[17] == '8') // avx512.mask.psrav8.si
@ -1494,8 +1501,9 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
IID = IsImmediate ? Intrinsic::x86_sse2_psrai_d
: Intrinsic::x86_sse2_psra_d;
else if (Size == 'q') // avx512.mask.psra.q.128, avx512.mask.psra.qi.128
IID = IsImmediate ? Intrinsic::x86_avx512_psrai_q_128
: Intrinsic::x86_avx512_psra_q_128;
IID = IsImmediate ? Intrinsic::x86_avx512_psrai_q_128 :
IsVariable ? Intrinsic::x86_avx512_psrav_q_128 :
Intrinsic::x86_avx512_psra_q_128;
else if (Size == 'w') // avx512.mask.psra.w.128, avx512.mask.psra.wi.128
IID = IsImmediate ? Intrinsic::x86_sse2_psrai_w
: Intrinsic::x86_sse2_psra_w;
@ -1506,20 +1514,23 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) {
IID = IsImmediate ? Intrinsic::x86_avx2_psrai_d
: Intrinsic::x86_avx2_psra_d;
else if (Size == 'q') // avx512.mask.psra.q.256, avx512.mask.psra.qi.256
IID = IsImmediate ? Intrinsic::x86_avx512_psrai_q_256
: Intrinsic::x86_avx512_psra_q_256;
IID = IsImmediate ? Intrinsic::x86_avx512_psrai_q_256 :
IsVariable ? Intrinsic::x86_avx512_psrav_q_256 :
Intrinsic::x86_avx512_psra_q_256;
else if (Size == 'w') // avx512.mask.psra.w.256, avx512.mask.psra.wi.256
IID = IsImmediate ? Intrinsic::x86_avx2_psrai_w
: Intrinsic::x86_avx2_psra_w;
else
llvm_unreachable("Unexpected size");
} else {
if (Size == 'd') // psra.di.512, psrai.d, psra.d
IID = IsImmediate ? Intrinsic::x86_avx512_psrai_d_512
: Intrinsic::x86_avx512_psra_d_512;
if (Size == 'd') // psra.di.512, psrai.d, psra.d, psrav.d.512
IID = IsImmediate ? Intrinsic::x86_avx512_psrai_d_512 :
IsVariable ? Intrinsic::x86_avx512_psrav_d_512 :
Intrinsic::x86_avx512_psra_d_512;
else if (Size == 'q') // psra.qi.512, psrai.q, psra.q
IID = IsImmediate ? Intrinsic::x86_avx512_psrai_q_512
: Intrinsic::x86_avx512_psra_q_512;
IID = IsImmediate ? Intrinsic::x86_avx512_psrai_q_512 :
IsVariable ? Intrinsic::x86_avx512_psrav_q_512 :
Intrinsic::x86_avx512_psra_q_512;
else if (Size == 'w') // psra.wi.512, psrai.w, psra.w
IID = IsImmediate ? Intrinsic::x86_avx512_psrai_w_512
: Intrinsic::x86_avx512_psra_w_512;

8
lib/Target/X86/X86IntrinsicsInfo.h
View File

@ -1073,20 +1073,12 @@ static const IntrinsicData IntrinsicsWithoutChain[] = {
X86_INTRINSIC_DATA(avx512_mask_prorv_q_512, INTR_TYPE_2OP_MASK, ISD::ROTR, 0),
X86_INTRINSIC_DATA(avx512_mask_pshuf_b_512, INTR_TYPE_2OP_MASK,
X86ISD::PSHUFB, 0),
X86_INTRINSIC_DATA(avx512_mask_psllv_d, INTR_TYPE_2OP_MASK, ISD::SHL, 0),
X86_INTRINSIC_DATA(avx512_mask_psllv_q, INTR_TYPE_2OP_MASK, ISD::SHL, 0),
X86_INTRINSIC_DATA(avx512_mask_psllv16_hi, INTR_TYPE_2OP_MASK, ISD::SHL, 0),
X86_INTRINSIC_DATA(avx512_mask_psllv32hi, INTR_TYPE_2OP_MASK, ISD::SHL, 0),
X86_INTRINSIC_DATA(avx512_mask_psllv8_hi, INTR_TYPE_2OP_MASK, ISD::SHL, 0),
X86_INTRINSIC_DATA(avx512_mask_psrav_d, INTR_TYPE_2OP_MASK, X86ISD::VSRAV, 0),
X86_INTRINSIC_DATA(avx512_mask_psrav_q, INTR_TYPE_2OP_MASK, X86ISD::VSRAV, 0),
X86_INTRINSIC_DATA(avx512_mask_psrav_q_128, INTR_TYPE_2OP_MASK, X86ISD::VSRAV, 0),
X86_INTRINSIC_DATA(avx512_mask_psrav_q_256, INTR_TYPE_2OP_MASK, X86ISD::VSRAV, 0),
X86_INTRINSIC_DATA(avx512_mask_psrav16_hi, INTR_TYPE_2OP_MASK, X86ISD::VSRAV, 0),
X86_INTRINSIC_DATA(avx512_mask_psrav32_hi, INTR_TYPE_2OP_MASK, X86ISD::VSRAV, 0),
X86_INTRINSIC_DATA(avx512_mask_psrav8_hi, INTR_TYPE_2OP_MASK, X86ISD::VSRAV, 0),
X86_INTRINSIC_DATA(avx512_mask_psrlv_d, INTR_TYPE_2OP_MASK, ISD::SRL, 0),
X86_INTRINSIC_DATA(avx512_mask_psrlv_q, INTR_TYPE_2OP_MASK, ISD::SRL, 0),
X86_INTRINSIC_DATA(avx512_mask_psrlv16_hi, INTR_TYPE_2OP_MASK, ISD::SRL, 0),
X86_INTRINSIC_DATA(avx512_mask_psrlv32hi, INTR_TYPE_2OP_MASK, ISD::SRL, 0),
X86_INTRINSIC_DATA(avx512_mask_psrlv8_hi, INTR_TYPE_2OP_MASK, ISD::SRL, 0),

203
test/CodeGen/X86/avx512-intrinsics-upgrade.ll
View File

@ -2279,3 +2279,206 @@ define <8 x i64> @test_x86_avx512_maskz_psra_q(<8 x i64> %a0, <2 x i64> %a1, i8
declare <8 x i64> @llvm.x86.avx512.mask.psra.q(<8 x i64>, <2 x i64>, <8 x i64>, i8) nounwind readnone
define <16 x i32> @test_x86_avx512_psllv_d(<16 x i32> %a0, <16 x i32> %a1) {
; CHECK-LABEL: test_x86_avx512_psllv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsllvd %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psllv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 -1)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_mask_psllv_d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psllv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsllvd %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psllv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_maskz_psllv_d(<16 x i32> %a0, <16 x i32> %a1, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psllv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsllvd %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psllv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 %mask)
ret <16 x i32> %res
}
declare <16 x i32> @llvm.x86.avx512.mask.psllv.d(<16 x i32>, <16 x i32>, <16 x i32>, i16) nounwind readnone
define <8 x i64> @test_x86_avx512_psllv_q(<8 x i64> %a0, <8 x i64> %a1) {
; CHECK-LABEL: test_x86_avx512_psllv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsllvq %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psllv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 -1)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_mask_psllv_q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psllv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsllvq %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psllv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_maskz_psllv_q(<8 x i64> %a0, <8 x i64> %a1, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psllv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsllvq %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psllv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 %mask)
ret <8 x i64> %res
}
declare <8 x i64> @llvm.x86.avx512.mask.psllv.q(<8 x i64>, <8 x i64>, <8 x i64>, i8) nounwind readnone
define <16 x i32> @test_x86_avx512_psrav_d(<16 x i32> %a0, <16 x i32> %a1) {
; CHECK-LABEL: test_x86_avx512_psrav_d:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsravd %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrav.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 -1)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_mask_psrav_d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psrav_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsravd %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrav.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_maskz_psrav_d(<16 x i32> %a0, <16 x i32> %a1, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psrav_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsravd %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrav.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 %mask)
ret <16 x i32> %res
}
declare <16 x i32> @llvm.x86.avx512.mask.psrav.d(<16 x i32>, <16 x i32>, <16 x i32>, i16) nounwind readnone
define <8 x i64> @test_x86_avx512_psrav_q(<8 x i64> %a0, <8 x i64> %a1) {
; CHECK-LABEL: test_x86_avx512_psrav_q:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsravq %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrav.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 -1)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_mask_psrav_q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psrav_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsravq %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrav.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_maskz_psrav_q(<8 x i64> %a0, <8 x i64> %a1, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psrav_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsravq %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrav.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 %mask)
ret <8 x i64> %res
}
declare <8 x i64> @llvm.x86.avx512.mask.psrav.q(<8 x i64>, <8 x i64>, <8 x i64>, i8) nounwind readnone
define <16 x i32> @test_x86_avx512_psrlv_d(<16 x i32> %a0, <16 x i32> %a1) {
; CHECK-LABEL: test_x86_avx512_psrlv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsrlvd %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrlv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 -1)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_mask_psrlv_d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psrlv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsrlvd %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrlv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_maskz_psrlv_d(<16 x i32> %a0, <16 x i32> %a1, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psrlv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsrlvd %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrlv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 %mask)
ret <16 x i32> %res
}
declare <16 x i32> @llvm.x86.avx512.mask.psrlv.d(<16 x i32>, <16 x i32>, <16 x i32>, i16) nounwind readnone
define <8 x i64> @test_x86_avx512_psrlv_q(<8 x i64> %a0, <8 x i64> %a1) {
; CHECK-LABEL: test_x86_avx512_psrlv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsrlvq %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrlv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 -1)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_mask_psrlv_q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psrlv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsrlvq %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrlv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_maskz_psrlv_q(<8 x i64> %a0, <8 x i64> %a1, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psrlv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsrlvq %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrlv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 %mask)
ret <8 x i64> %res
}
declare <8 x i64> @llvm.x86.avx512.mask.psrlv.q(<8 x i64>, <8 x i64>, <8 x i64>, i8) nounwind readnone
define <8 x i64> @test_x86_avx512_psrlv_q_memop(<8 x i64> %a0, <8 x i64>* %ptr) {
; CHECK-LABEL: test_x86_avx512_psrlv_q_memop:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsrlvq (%rdi), %zmm0, %zmm0
; CHECK-NEXT: retq
%b = load <8 x i64>, <8 x i64>* %ptr
%res = call <8 x i64> @llvm.x86.avx512.mask.psrlv.q(<8 x i64> %a0, <8 x i64> %b, <8 x i64> zeroinitializer, i8 -1)
ret <8 x i64> %res
}

203
test/CodeGen/X86/avx512-intrinsics.ll
View File

@ -1324,209 +1324,6 @@ define <4 x double> @test_vextractf64x4(<8 x double> %a) {
declare <4 x double> @llvm.x86.avx512.mask.vextractf64x4.512(<8 x double>, i32, <4 x double>, i8)
define <16 x i32> @test_x86_avx512_psllv_d(<16 x i32> %a0, <16 x i32> %a1) {
; CHECK-LABEL: test_x86_avx512_psllv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsllvd %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psllv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 -1)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_mask_psllv_d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psllv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsllvd %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psllv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_maskz_psllv_d(<16 x i32> %a0, <16 x i32> %a1, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psllv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsllvd %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psllv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 %mask)
ret <16 x i32> %res
}
declare <16 x i32> @llvm.x86.avx512.mask.psllv.d(<16 x i32>, <16 x i32>, <16 x i32>, i16) nounwind readnone
define <8 x i64> @test_x86_avx512_psllv_q(<8 x i64> %a0, <8 x i64> %a1) {
; CHECK-LABEL: test_x86_avx512_psllv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsllvq %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psllv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 -1)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_mask_psllv_q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psllv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsllvq %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psllv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_maskz_psllv_q(<8 x i64> %a0, <8 x i64> %a1, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psllv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsllvq %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psllv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 %mask)
ret <8 x i64> %res
}
declare <8 x i64> @llvm.x86.avx512.mask.psllv.q(<8 x i64>, <8 x i64>, <8 x i64>, i8) nounwind readnone
define <16 x i32> @test_x86_avx512_psrav_d(<16 x i32> %a0, <16 x i32> %a1) {
; CHECK-LABEL: test_x86_avx512_psrav_d:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsravd %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrav.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 -1)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_mask_psrav_d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psrav_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsravd %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrav.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_maskz_psrav_d(<16 x i32> %a0, <16 x i32> %a1, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psrav_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsravd %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrav.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 %mask)
ret <16 x i32> %res
}
declare <16 x i32> @llvm.x86.avx512.mask.psrav.d(<16 x i32>, <16 x i32>, <16 x i32>, i16) nounwind readnone
define <8 x i64> @test_x86_avx512_psrav_q(<8 x i64> %a0, <8 x i64> %a1) {
; CHECK-LABEL: test_x86_avx512_psrav_q:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsravq %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrav.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 -1)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_mask_psrav_q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psrav_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsravq %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrav.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_maskz_psrav_q(<8 x i64> %a0, <8 x i64> %a1, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psrav_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsravq %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrav.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 %mask)
ret <8 x i64> %res
}
declare <8 x i64> @llvm.x86.avx512.mask.psrav.q(<8 x i64>, <8 x i64>, <8 x i64>, i8) nounwind readnone
define <16 x i32> @test_x86_avx512_psrlv_d(<16 x i32> %a0, <16 x i32> %a1) {
; CHECK-LABEL: test_x86_avx512_psrlv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsrlvd %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrlv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 -1)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_mask_psrlv_d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psrlv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsrlvd %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrlv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> %a2, i16 %mask)
ret <16 x i32> %res
}
define <16 x i32> @test_x86_avx512_maskz_psrlv_d(<16 x i32> %a0, <16 x i32> %a1, i16 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psrlv_d:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsrlvd %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <16 x i32> @llvm.x86.avx512.mask.psrlv.d(<16 x i32> %a0, <16 x i32> %a1, <16 x i32> zeroinitializer, i16 %mask)
ret <16 x i32> %res
}
declare <16 x i32> @llvm.x86.avx512.mask.psrlv.d(<16 x i32>, <16 x i32>, <16 x i32>, i16) nounwind readnone
define <8 x i64> @test_x86_avx512_psrlv_q(<8 x i64> %a0, <8 x i64> %a1) {
; CHECK-LABEL: test_x86_avx512_psrlv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsrlvq %zmm1, %zmm0, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrlv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 -1)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_mask_psrlv_q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_mask_psrlv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsrlvq %zmm1, %zmm0, %zmm2 {%k1}
; CHECK-NEXT: vmovdqa64 %zmm2, %zmm0
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrlv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> %a2, i8 %mask)
ret <8 x i64> %res
}
define <8 x i64> @test_x86_avx512_maskz_psrlv_q(<8 x i64> %a0, <8 x i64> %a1, i8 %mask) {
; CHECK-LABEL: test_x86_avx512_maskz_psrlv_q:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1
; CHECK-NEXT: vpsrlvq %zmm1, %zmm0, %zmm0 {%k1} {z}
; CHECK-NEXT: retq
%res = call <8 x i64> @llvm.x86.avx512.mask.psrlv.q(<8 x i64> %a0, <8 x i64> %a1, <8 x i64> zeroinitializer, i8 %mask)
ret <8 x i64> %res
}
declare <8 x i64> @llvm.x86.avx512.mask.psrlv.q(<8 x i64>, <8 x i64>, <8 x i64>, i8) nounwind readnone
define <8 x i64> @test_x86_avx512_psrlv_q_memop(<8 x i64> %a0, <8 x i64>* %ptr) {
; CHECK-LABEL: test_x86_avx512_psrlv_q_memop:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsrlvq (%rdi), %zmm0, %zmm0
; CHECK-NEXT: retq
%b = load <8 x i64>, <8 x i64>* %ptr
%res = call <8 x i64> @llvm.x86.avx512.mask.psrlv.q(<8 x i64> %a0, <8 x i64> %b, <8 x i64> zeroinitializer, i8 -1)
ret <8 x i64> %res
}
declare <16 x float> @llvm.x86.avx512.mask.sub.ps.512(<16 x float>, <16 x float>, <16 x float>, i16, i32)
declare <16 x float> @llvm.x86.avx512.mask.mul.ps.512(<16 x float>, <16 x float>, <16 x float>, i16, i32)
declare <8 x double> @llvm.x86.avx512.mask.mul.pd.512(<8 x double>, <8 x double>, <8 x double>, i8, i32)

53
test/CodeGen/X86/avx512vl-intrinsics-upgrade.ll
View File

@ -4499,3 +4499,56 @@ define <4 x i64>@test_int_x86_avx512_mask_psra_qi_256(<4 x i64> %x0, i32 %x1, <4
ret <4 x i64> %res4
}
declare <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64>, <2 x i64>, <2 x i64>, i8)
define <2 x i64>@test_int_x86_avx512_mask_psrav_q_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_mask_psrav_q_128:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsravq %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf2,0xfd,0x08,0x46,0xd9]
; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf]
; CHECK-NEXT: vpsravq %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf2,0xfd,0x09,0x46,0xd1]
; CHECK-NEXT: vpsravq %xmm1, %xmm0, %xmm0 {%k1} {z} ## encoding: [0x62,0xf2,0xfd,0x89,0x46,0xc1]
; CHECK-NEXT: vpaddq %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0xed,0x08,0xd4,0xc0]
; CHECK-NEXT: vpaddq %xmm3, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0xfd,0x08,0xd4,0xc3]
; CHECK-NEXT: retq ## encoding: [0xc3]
%res = call <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3)
%res1 = call <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3)
%res2 = call <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1)
%res3 = add <2 x i64> %res, %res1
%res4 = add <2 x i64> %res3, %res2
ret <2 x i64> %res4
}
define <2 x i64>@test_int_x86_avx512_mask_psrav_q_128_const(i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_mask_psrav_q_128_const:
; CHECK: ## BB#0:
; CHECK-NEXT: vmovdqa64 {{.*#+}} xmm0 = [2,18446744073709551607]
; CHECK-NEXT: ## encoding: [0x62,0xf1,0xfd,0x08,0x6f,0x05,A,A,A,A]
; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI304_0-4, kind: reloc_riprel_4byte
; CHECK-NEXT: vpsravq {{.*}}(%rip), %xmm0, %xmm0 ## encoding: [0x62,0xf2,0xfd,0x08,0x46,0x05,A,A,A,A]
; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI304_1-4, kind: reloc_riprel_4byte
; CHECK-NEXT: retq ## encoding: [0xc3]
%res = call <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64> <i64 2, i64 -9>, <2 x i64> <i64 1, i64 90>, <2 x i64> zeroinitializer, i8 -1)
ret <2 x i64> %res
}
declare <4 x i64> @llvm.x86.avx512.mask.psrav.q.256(<4 x i64>, <4 x i64>, <4 x i64>, i8)
define <4 x i64>@test_int_x86_avx512_mask_psrav_q_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_mask_psrav_q_256:
; CHECK: ## BB#0:
; CHECK-NEXT: vpsravq %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf2,0xfd,0x28,0x46,0xd9]
; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf]
; CHECK-NEXT: vpsravq %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf2,0xfd,0x29,0x46,0xd1]
; CHECK-NEXT: vpsravq %ymm1, %ymm0, %ymm0 {%k1} {z} ## encoding: [0x62,0xf2,0xfd,0xa9,0x46,0xc1]
; CHECK-NEXT: vpaddq %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0xed,0x28,0xd4,0xc0]
; CHECK-NEXT: vpaddq %ymm3, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0xfd,0x28,0xd4,0xc3]
; CHECK-NEXT: retq ## encoding: [0xc3]
%res = call <4 x i64> @llvm.x86.avx512.mask.psrav.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3)
%res1 = call <4 x i64> @llvm.x86.avx512.mask.psrav.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
%res2 = call <4 x i64> @llvm.x86.avx512.mask.psrav.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1)
%res3 = add <4 x i64> %res, %res1
%res4 = add <4 x i64> %res3, %res2
ret <4 x i64> %res4
}

53
test/CodeGen/X86/avx512vl-intrinsics.ll
View File

@ -4422,59 +4422,6 @@ define <8 x i32>@test_int_x86_avx512_mask_broadcasti32x4_256(<4 x i32> %x0, <8 x
ret <8 x i32> %res5
}
declare <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64>, <2 x i64>, <2 x i64>, i8)
define <2 x i64>@test_int_x86_avx512_mask_psrav_q_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_mask_psrav_q_128:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf]
; CHECK-NEXT: vpsravq %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf2,0xfd,0x09,0x46,0xd1]
; CHECK-NEXT: vpsravq %xmm1, %xmm0, %xmm3 {%k1} {z} ## encoding: [0x62,0xf2,0xfd,0x89,0x46,0xd9]
; CHECK-NEXT: vpsravq %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf2,0xfd,0x08,0x46,0xc1]
; CHECK-NEXT: vpaddq %xmm3, %xmm2, %xmm1 ## encoding: [0x62,0xf1,0xed,0x08,0xd4,0xcb]
; CHECK-NEXT: vpaddq %xmm0, %xmm1, %xmm0 ## encoding: [0x62,0xf1,0xf5,0x08,0xd4,0xc0]
; CHECK-NEXT: retq ## encoding: [0xc3]
%res = call <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3)
%res1 = call <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3)
%res2 = call <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1)
%res3 = add <2 x i64> %res, %res1
%res4 = add <2 x i64> %res3, %res2
ret <2 x i64> %res4
}
define <2 x i64>@test_int_x86_avx512_mask_psrav_q_128_const(i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_mask_psrav_q_128_const:
; CHECK: ## BB#0:
; CHECK-NEXT: vmovdqa64 {{.*#+}} xmm0 = [2,18446744073709551607]
; CHECK-NEXT: ## encoding: [0x62,0xf1,0xfd,0x08,0x6f,0x05,A,A,A,A]
; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI266_0-4, kind: reloc_riprel_4byte
; CHECK-NEXT: vpsravq {{.*}}(%rip), %xmm0, %xmm0 ## encoding: [0x62,0xf2,0xfd,0x08,0x46,0x05,A,A,A,A]
; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI266_1-4, kind: reloc_riprel_4byte
; CHECK-NEXT: retq ## encoding: [0xc3]
%res = call <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64> <i64 2, i64 -9>, <2 x i64> <i64 1, i64 90>, <2 x i64> zeroinitializer, i8 -1)
ret <2 x i64> %res
}
declare <4 x i64> @llvm.x86.avx512.mask.psrav.q.256(<4 x i64>, <4 x i64>, <4 x i64>, i8)
define <4 x i64>@test_int_x86_avx512_mask_psrav_q_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) {
; CHECK-LABEL: test_int_x86_avx512_mask_psrav_q_256:
; CHECK: ## BB#0:
; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf]
; CHECK-NEXT: vpsravq %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf2,0xfd,0x29,0x46,0xd1]
; CHECK-NEXT: vpsravq %ymm1, %ymm0, %ymm3 {%k1} {z} ## encoding: [0x62,0xf2,0xfd,0xa9,0x46,0xd9]
; CHECK-NEXT: vpsravq %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf2,0xfd,0x28,0x46,0xc1]
; CHECK-NEXT: vpaddq %ymm3, %ymm2, %ymm1 ## encoding: [0x62,0xf1,0xed,0x28,0xd4,0xcb]
; CHECK-NEXT: vpaddq %ymm0, %ymm1, %ymm0 ## encoding: [0x62,0xf1,0xf5,0x28,0xd4,0xc0]
; CHECK-NEXT: retq ## encoding: [0xc3]
%res = call <4 x i64> @llvm.x86.avx512.mask.psrav.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3)
%res1 = call <4 x i64> @llvm.x86.avx512.mask.psrav.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3)
%res2 = call <4 x i64> @llvm.x86.avx512.mask.psrav.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1)
%res3 = add <4 x i64> %res, %res1
%res4 = add <4 x i64> %res3, %res2
ret <4 x i64> %res4
}
declare <4 x i32> @llvm.x86.avx512.mask.prorv.d.128(<4 x i32>, <4 x i32>, <4 x i32>, i8)
define <4 x i32>@test_int_x86_avx512_mask_prorv_d_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) {