RPCSX/llvm
1
0
Fork 0
mirror of https://github.com/RPCSX/llvm.git synced 2024-12-11 13:46:13 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

[InstCombine][AVX2] Added VPERMD/VPERMPS shuffle combining placeholder tests.

Browse Source 
For future support for VPERMD/VPERMPS to generic shuffles combines

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@268166 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Simon Pilgrim 2016-04-30 20:41:52 +00:00
parent 5ac1e61a27
commit 585ee3d03c
1 changed files with 87 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

87
test/Transforms/InstCombine/x86-avx2.ll Normal file
View File

@ -0,0 +1,87 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -instcombine -S | FileCheck %s
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
; FIXME: Verify that instcombine is able to fold identity shuffles.
define <8 x i32> @identity_test_vpermd(<8 x i32> %a0) {
; CHECK-LABEL: @identity_test_vpermd(
; CHECK-NEXT: [[A:%.*]] = tail call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> [[A:%.*]]0, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>)
; CHECK-NEXT: ret <8 x i32> [[A]]
;
%a = tail call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> %a0, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>)
ret <8 x i32> %a
}
define <8 x float> @identity_test_vpermps(<8 x float> %a0) {
; CHECK-LABEL: @identity_test_vpermps(
; CHECK-NEXT: [[A:%.*]] = tail call <8 x float> @llvm.x86.avx2.permps(<8 x float> [[A:%.*]]0, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>)
; CHECK-NEXT: ret <8 x float> [[A]]
;
%a = tail call <8 x float> @llvm.x86.avx2.permps(<8 x float> %a0, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>)
ret <8 x float> %a
}
; FIXME: Instcombine should be able to fold the following shuffle to a builtin shufflevector
; with a shuffle mask of all zeroes.
define <8 x i32> @zero_test_vpermd(<8 x i32> %a0) {
; CHECK-LABEL: @zero_test_vpermd(
; CHECK-NEXT: [[A:%.*]] = tail call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> [[A:%.*]]0, <8 x i32> zeroinitializer)
; CHECK-NEXT: ret <8 x i32> [[A]]
;
%a = tail call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> %a0, <8 x i32> zeroinitializer)
ret <8 x i32> %a
}
define <8 x float> @zero_test_vpermps(<8 x float> %a0) {
; CHECK-LABEL: @zero_test_vpermps(
; CHECK-NEXT: [[A:%.*]] = tail call <8 x float> @llvm.x86.avx2.permps(<8 x float> [[A:%.*]]0, <8 x i32> zeroinitializer)
; CHECK-NEXT: ret <8 x float> [[A]]
;
%a = tail call <8 x float> @llvm.x86.avx2.permps(<8 x float> %a0, <8 x i32> zeroinitializer)
ret <8 x float> %a
}
; FIXME: Verify that instcombine is able to fold constant shuffles.
define <8 x i32> @shuffle_test_vpermd(<8 x i32> %a0) {
; CHECK-LABEL: @shuffle_test_vpermd(
; CHECK-NEXT: [[A:%.*]] = tail call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> [[A:%.*]]0, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>)
; CHECK-NEXT: ret <8 x i32> [[A]]
;
%a = tail call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> %a0, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>)
ret <8 x i32> %a
}
define <8 x float> @shuffle_test_vpermps(<8 x float> %a0) {
; CHECK-LABEL: @shuffle_test_vpermps(
; CHECK-NEXT: [[A:%.*]] = tail call <8 x float> @llvm.x86.avx2.permps(<8 x float> [[A:%.*]]0, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>)
; CHECK-NEXT: ret <8 x float> [[A]]
;
%a = tail call <8 x float> @llvm.x86.avx2.permps(<8 x float> %a0, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>)
ret <8 x float> %a
}
; FIXME: Verify that instcombine is able to fold constant shuffles with undef mask elements.
define <8 x i32> @undef_test_vpermd(<8 x i32> %a0) {
; CHECK-LABEL: @undef_test_vpermd(
; CHECK-NEXT: [[A:%.*]] = tail call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> [[A:%.*]]0, <8 x i32> <i32 undef, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>)
; CHECK-NEXT: ret <8 x i32> [[A]]
;
%a = tail call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> %a0, <8 x i32> <i32 undef, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>)
ret <8 x i32> %a
}
define <8 x float> @undef_test_vpermps(<8 x float> %a0) {
; CHECK-LABEL: @undef_test_vpermps(
; CHECK-NEXT: [[A:%.*]] = tail call <8 x float> @llvm.x86.avx2.permps(<8 x float> [[A:%.*]]0, <8 x i32> <i32 undef, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>)
; CHECK-NEXT: ret <8 x float> [[A]]
;
%a = tail call <8 x float> @llvm.x86.avx2.permps(<8 x float> %a0, <8 x i32> <i32 undef, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>)
ret <8 x float> %a
}
declare <8 x i32> @llvm.x86.avx2.permd(<8 x i32>, <8 x i32>)
declare <8 x float> @llvm.x86.avx2.permps(<8 x float>, <8 x i32>)