maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[PowerPC] Don't use xscvdpspn on the P7 

xscvdpspn was not introduced until the P8, so don't use it on the P7. Fixes a
regression introduced in r288152.

llvm-svn: 312612
This commit is contained in:
Hal Finkel 2017-09-06 03:08:26 +00:00
parent a69a3a3f62
commit 112a6bac72
2 changed files with 33 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
llvm/lib/Target/PowerPC/PPCISelLowering.cpp
View File

@ -7463,9 +7463,11 @@ static SDValue BuildVSLDOI(SDValue LHS, SDValue RHS, unsigned Amt, EVT VT,
/// - The node is a "load-and-splat"
/// In all other cases, we will choose to keep the BUILD_VECTOR.
static bool haveEfficientBuildVectorPattern(BuildVectorSDNode *V,
bool HasDirectMove) {
bool HasDirectMove,
bool HasP8Vector) {
EVT VecVT = V->getValueType(0);
bool RightType = VecVT == MVT::v2f64 || VecVT == MVT::v4f32 ||
bool RightType = VecVT == MVT::v2f64 ||
(HasP8Vector && VecVT == MVT::v4f32) ||
(HasDirectMove && (VecVT == MVT::v2i64 || VecVT == MVT::v4i32));
if (!RightType)
return false;
@ -7627,7 +7629,8 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
// lowered to VSX instructions under certain conditions.
// Without VSX, there is no pattern more efficient than expanding the node.
if (Subtarget.hasVSX() &&
haveEfficientBuildVectorPattern(BVN, Subtarget.hasDirectMove()))
haveEfficientBuildVectorPattern(BVN, Subtarget.hasDirectMove(),
Subtarget.hasP8Vector()))
return Op;
return SDValue();
}

27
llvm/test/CodeGen/PowerPC/fp-splat.ll Normal file
View File

@ -0,0 +1,27 @@
; RUN: llc -mcpu=pwr8 -mtriple=powerpc64le-unknown-linux-gnu < %s | FileCheck %s -check-prefix=CHECK-P8 -check-prefix=CHECK
; RUN: llc -mcpu=pwr7 -mtriple=powerpc64-unknown-linux-gnu < %s | FileCheck %s -check-prefix=CHECK-P7 -check-prefix=CHECK
define <4 x float> @test1(float %a) {
entry:
; CHECK-LABEL: test1
%vecins = insertelement <4 x float> undef, float %a, i32 0
%vecins1 = insertelement <4 x float> %vecins, float %a, i32 1
%vecins2 = insertelement <4 x float> %vecins1, float %a, i32 2
%vecins3 = insertelement <4 x float> %vecins2, float %a, i32 3
ret <4 x float> %vecins3
; CHECK-P8: xscvdpspn
; CHECK-P7-NOT: xscvdpspn
; CHECK: blr
}
define <2 x double> @test2(double %a) {
entry:
; CHECK-LABEL: test2
%vecins = insertelement <2 x double> undef, double %a, i32 0
%vecins1 = insertelement <2 x double> %vecins, double %a, i32 1
ret <2 x double> %vecins1
; CHECK-P8: xxspltd
; CHECK-P7: xxspltd
; CHECK: blr
}