mirror of
https://github.com/RPCSX/llvm.git
synced 2024-12-15 07:59:50 +00:00
538287dea2
The ABI allows sub-128 vectors to be passed and returned in registers, with the vector occupying the upper part of a register. We therefore want to legalize those types by widening the vector rather than promoting the elements. The patch includes some simple tests for sub-128 vectors and also tests that we can recognize various pack sequences, some of which use sub-128 vectors as temporary results. One of these forms is based on the pack sequences generated by llvmpipe when no intrinsics are used. Signed unpacks are recognized as BUILD_VECTORs whose elements are individually sign-extended. Unsigned unpacks can have the equivalent form with zero extension, but they also occur as shuffles in which some elements are zero. Based on a patch by Richard Sandiford. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@236525 91177308-0d34-0410-b5e6-96231b3b80d8
156 lines
5.5 KiB
LLVM
156 lines
5.5 KiB
LLVM
; Test various target-specific DAG combiner patterns.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 | FileCheck %s
|
|
|
|
; Check that an extraction followed by a truncation is effectively treated
|
|
; as a bitcast.
|
|
define void @f1(<4 x i32> %v1, <4 x i32> %v2, i8 *%ptr1, i8 *%ptr2) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK: vaf [[REG:%v[0-9]+]], %v24, %v26
|
|
; CHECK-DAG: vsteb [[REG]], 0(%r2), 3
|
|
; CHECK-DAG: vsteb [[REG]], 0(%r3), 15
|
|
; CHECK: br %r14
|
|
%add = add <4 x i32> %v1, %v2
|
|
%elem1 = extractelement <4 x i32> %add, i32 0
|
|
%elem2 = extractelement <4 x i32> %add, i32 3
|
|
%trunc1 = trunc i32 %elem1 to i8
|
|
%trunc2 = trunc i32 %elem2 to i8
|
|
store i8 %trunc1, i8 *%ptr1
|
|
store i8 %trunc2, i8 *%ptr2
|
|
ret void
|
|
}
|
|
|
|
; Test a case where a pack-type shuffle can be eliminated.
|
|
define i16 @f2(<4 x i32> %v1, <4 x i32> %v2, <4 x i32> %v3) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK-NOT: vpk
|
|
; CHECK-DAG: vaf [[REG1:%v[0-9]+]], %v24, %v26
|
|
; CHECK-DAG: vaf [[REG2:%v[0-9]+]], %v26, %v28
|
|
; CHECK-DAG: vlgvh {{%r[0-5]}}, [[REG1]], 3
|
|
; CHECK-DAG: vlgvh {{%r[0-5]}}, [[REG2]], 7
|
|
; CHECK: br %r14
|
|
%add1 = add <4 x i32> %v1, %v2
|
|
%add2 = add <4 x i32> %v2, %v3
|
|
%shuffle = shufflevector <4 x i32> %add1, <4 x i32> %add2,
|
|
<4 x i32> <i32 1, i32 3, i32 5, i32 7>
|
|
%bitcast = bitcast <4 x i32> %shuffle to <8 x i16>
|
|
%elem1 = extractelement <8 x i16> %bitcast, i32 1
|
|
%elem2 = extractelement <8 x i16> %bitcast, i32 7
|
|
%res = add i16 %elem1, %elem2
|
|
ret i16 %res
|
|
}
|
|
|
|
; ...and again in a case where there's also a splat and a bitcast.
|
|
define i16 @f3(<4 x i32> %v1, <4 x i32> %v2, <2 x i64> %v3) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK-NOT: vrepg
|
|
; CHECK-NOT: vpk
|
|
; CHECK-DAG: vaf [[REG:%v[0-9]+]], %v24, %v26
|
|
; CHECK-DAG: vlgvh {{%r[0-5]}}, [[REG]], 6
|
|
; CHECK-DAG: vlgvh {{%r[0-5]}}, %v28, 3
|
|
; CHECK: br %r14
|
|
%add = add <4 x i32> %v1, %v2
|
|
%splat = shufflevector <2 x i64> %v3, <2 x i64> undef,
|
|
<2 x i32> <i32 0, i32 0>
|
|
%splatcast = bitcast <2 x i64> %splat to <4 x i32>
|
|
%shuffle = shufflevector <4 x i32> %add, <4 x i32> %splatcast,
|
|
<4 x i32> <i32 1, i32 3, i32 5, i32 7>
|
|
%bitcast = bitcast <4 x i32> %shuffle to <8 x i16>
|
|
%elem1 = extractelement <8 x i16> %bitcast, i32 2
|
|
%elem2 = extractelement <8 x i16> %bitcast, i32 7
|
|
%res = add i16 %elem1, %elem2
|
|
ret i16 %res
|
|
}
|
|
|
|
; ...and again with a merge low instead of a pack.
|
|
define i16 @f4(<4 x i32> %v1, <4 x i32> %v2, <2 x i64> %v3) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK-NOT: vrepg
|
|
; CHECK-NOT: vmr
|
|
; CHECK-DAG: vaf [[REG:%v[0-9]+]], %v24, %v26
|
|
; CHECK-DAG: vlgvh {{%r[0-5]}}, [[REG]], 6
|
|
; CHECK-DAG: vlgvh {{%r[0-5]}}, %v28, 3
|
|
; CHECK: br %r14
|
|
%add = add <4 x i32> %v1, %v2
|
|
%splat = shufflevector <2 x i64> %v3, <2 x i64> undef,
|
|
<2 x i32> <i32 0, i32 0>
|
|
%splatcast = bitcast <2 x i64> %splat to <4 x i32>
|
|
%shuffle = shufflevector <4 x i32> %add, <4 x i32> %splatcast,
|
|
<4 x i32> <i32 2, i32 6, i32 3, i32 7>
|
|
%bitcast = bitcast <4 x i32> %shuffle to <8 x i16>
|
|
%elem1 = extractelement <8 x i16> %bitcast, i32 4
|
|
%elem2 = extractelement <8 x i16> %bitcast, i32 7
|
|
%res = add i16 %elem1, %elem2
|
|
ret i16 %res
|
|
}
|
|
|
|
; ...and again with a merge high.
|
|
define i16 @f5(<4 x i32> %v1, <4 x i32> %v2, <2 x i64> %v3) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK-NOT: vrepg
|
|
; CHECK-NOT: vmr
|
|
; CHECK-DAG: vaf [[REG:%v[0-9]+]], %v24, %v26
|
|
; CHECK-DAG: vlgvh {{%r[0-5]}}, [[REG]], 2
|
|
; CHECK-DAG: vlgvh {{%r[0-5]}}, %v28, 3
|
|
; CHECK: br %r14
|
|
%add = add <4 x i32> %v1, %v2
|
|
%splat = shufflevector <2 x i64> %v3, <2 x i64> undef,
|
|
<2 x i32> <i32 0, i32 0>
|
|
%splatcast = bitcast <2 x i64> %splat to <4 x i32>
|
|
%shuffle = shufflevector <4 x i32> %add, <4 x i32> %splatcast,
|
|
<4 x i32> <i32 0, i32 4, i32 1, i32 5>
|
|
%bitcast = bitcast <4 x i32> %shuffle to <8 x i16>
|
|
%elem1 = extractelement <8 x i16> %bitcast, i32 4
|
|
%elem2 = extractelement <8 x i16> %bitcast, i32 7
|
|
%res = add i16 %elem1, %elem2
|
|
ret i16 %res
|
|
}
|
|
|
|
; Test a case where an unpack high can be eliminated from the usual
|
|
; load-extend sequence.
|
|
define void @f6(<8 x i8> *%ptr1, i8 *%ptr2, i8 *%ptr3, i8 *%ptr4) {
|
|
; CHECK-LABEL: f6:
|
|
; CHECK: vlrepg [[REG:%v[0-9]+]], 0(%r2)
|
|
; CHECK-NOT: vup
|
|
; CHECK-DAG: vsteb [[REG]], 0(%r3), 1
|
|
; CHECK-DAG: vsteb [[REG]], 0(%r4), 2
|
|
; CHECK-DAG: vsteb [[REG]], 0(%r5), 7
|
|
; CHECK: br %r14
|
|
%vec = load <8 x i8>, <8 x i8> *%ptr1
|
|
%ext = sext <8 x i8> %vec to <8 x i16>
|
|
%elem1 = extractelement <8 x i16> %ext, i32 1
|
|
%elem2 = extractelement <8 x i16> %ext, i32 2
|
|
%elem3 = extractelement <8 x i16> %ext, i32 7
|
|
%trunc1 = trunc i16 %elem1 to i8
|
|
%trunc2 = trunc i16 %elem2 to i8
|
|
%trunc3 = trunc i16 %elem3 to i8
|
|
store i8 %trunc1, i8 *%ptr2
|
|
store i8 %trunc2, i8 *%ptr3
|
|
store i8 %trunc3, i8 *%ptr4
|
|
ret void
|
|
}
|
|
|
|
; ...and again with a bitcast inbetween.
|
|
define void @f7(<4 x i8> *%ptr1, i8 *%ptr2, i8 *%ptr3, i8 *%ptr4) {
|
|
; CHECK-LABEL: f7:
|
|
; CHECK: vlrepf [[REG:%v[0-9]+]], 0(%r2)
|
|
; CHECK-NOT: vup
|
|
; CHECK-DAG: vsteb [[REG]], 0(%r3), 0
|
|
; CHECK-DAG: vsteb [[REG]], 0(%r4), 1
|
|
; CHECK-DAG: vsteb [[REG]], 0(%r5), 3
|
|
; CHECK: br %r14
|
|
%vec = load <4 x i8>, <4 x i8> *%ptr1
|
|
%ext = sext <4 x i8> %vec to <4 x i32>
|
|
%bitcast = bitcast <4 x i32> %ext to <8 x i16>
|
|
%elem1 = extractelement <8 x i16> %bitcast, i32 1
|
|
%elem2 = extractelement <8 x i16> %bitcast, i32 3
|
|
%elem3 = extractelement <8 x i16> %bitcast, i32 7
|
|
%trunc1 = trunc i16 %elem1 to i8
|
|
%trunc2 = trunc i16 %elem2 to i8
|
|
%trunc3 = trunc i16 %elem3 to i8
|
|
store i8 %trunc1, i8 *%ptr2
|
|
store i8 %trunc2, i8 *%ptr3
|
|
store i8 %trunc3, i8 *%ptr4
|
|
ret void
|
|
}
|