RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 04:02:41 +01:00
Code Issues Projects Releases Wiki Activity

Disable a couple more vector splat optimizations on PPC.

Browse Source 
I didn't see those because the test case used "not grep". FileCheck the test and
XFAIL it, preserving the old optimization, so this can be fixed eventually.

llvm-svn: 174330
This commit is contained in:
Benjamin Kramer 2013-02-04 15:52:32 +00:00
parent a75f1f87f7
commit d07b68b101
2 changed files with 45 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
lib/Target/PowerPC/PPCISelLowering.cpp
View File

@ -5138,14 +5138,16 @@ SDValue PPCTargetLowering::LowerBUILD_VECTOR(SDValue Op,
// Three instruction sequences.
// Odd, in range [17,31]: (vsplti C)-(vsplti -16).
if (SextVal >= 0 && SextVal <= 31) {
// FIXME: Disabled because the add gets constant folded.
if (0 && SextVal >= 0 && SextVal <= 31) {
SDValue LHS = BuildSplatI(SextVal-16, SplatSize, MVT::Other, DAG, dl);
SDValue RHS = BuildSplatI(-16, SplatSize, MVT::Other, DAG, dl);
LHS = DAG.getNode(ISD::SUB, dl, LHS.getValueType(), LHS, RHS);
return DAG.getNode(ISD::BITCAST, dl, Op.getValueType(), LHS);
}
// Odd, in range [-31,-17]: (vsplti C)+(vsplti -16).
if (SextVal >= -31 && SextVal <= 0) {
// FIXME: Disabled because the add gets constant folded.
if (0 && SextVal >= -31 && SextVal <= 0) {
SDValue LHS = BuildSplatI(SextVal+16, SplatSize, MVT::Other, DAG, dl);
SDValue RHS = BuildSplatI(-16, SplatSize, MVT::Other, DAG, dl);
LHS = DAG.getNode(ISD::ADD, dl, LHS.getValueType(), LHS, RHS);

42
test/CodeGen/PowerPC/vec_constants.ll
View File

@ -1,4 +1,5 @@
; RUN: llc < %s -march=ppc32 -mcpu=g5 | not grep CPI
; RUN: llc < %s -march=ppc32 -mcpu=g5 | FileCheck %s
; XFAIL: *
define void @test1(<4 x i32>* %P1, <4 x i32>* %P2, <4 x float>* %P3) nounwind {
%tmp = load <4 x i32>* %P1 ; <<4 x i32>> [#uses=1]
@ -13,32 +14,71 @@ define void @test1(<4 x i32>* %P1, <4 x i32>* %P2, <4 x float>* %P3) nounwind {
%tmp13 = bitcast <4 x i32> %tmp12 to <4 x float> ; <<4 x float>> [#uses=1]
store <4 x float> %tmp13, <4 x float>* %P3
ret void
; CHECK: test1:
; CHECK-NOT: CPI
}
define <4 x i32> @test_30() nounwind {
ret <4 x i32> < i32 30, i32 30, i32 30, i32 30 >
; CHECK: test_30:
; CHECK: vspltisw
; CHECK-NEXT: vadduwm
; CHECK-NEXT: blr
}
define <4 x i32> @test_29() nounwind {
ret <4 x i32> < i32 29, i32 29, i32 29, i32 29 >
; CHECK: test_29:
; CHECK: vspltisw
; CHECK-NEXT: vspltisw
; CHECK-NEXT: vsubuwm
; CHECK-NEXT: blr
}
define <8 x i16> @test_n30() nounwind {
ret <8 x i16> < i16 -30, i16 -30, i16 -30, i16 -30, i16 -30, i16 -30, i16 -30, i16 -30 >
; CHECK: test_n30:
; CHECK: vspltish
; CHECK-NEXT: vadduhm
; CHECK-NEXT: blr
}
define <16 x i8> @test_n104() nounwind {
ret <16 x i8> < i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104, i8 -104 >
; CHECK: test_n104:
; CHECK: vspltisb
; CHECK-NEXT: vslb
; CHECK-NEXT: blr
}
define <4 x i32> @test_vsldoi() nounwind {
ret <4 x i32> < i32 512, i32 512, i32 512, i32 512 >
; CHECK: test_vsldoi:
; CHECK: vspltisw
; CHECK-NEXT: vsldoi
; CHECK-NEXT: blr
}
define <8 x i16> @test_vsldoi_65023() nounwind {
ret <8 x i16> < i16 65023, i16 65023,i16 65023,i16 65023,i16 65023,i16 65023,i16 65023,i16 65023 >
; CHECK: test_vsldoi_65023:
; CHECK: vspltish
; CHECK-NEXT: vsldoi
; CHECK-NEXT: blr
}
define <4 x i32> @test_rol() nounwind {
ret <4 x i32> < i32 -11534337, i32 -11534337, i32 -11534337, i32 -11534337 >
; CHECK: test_rol:
; CHECK: vspltisw
; CHECK-NEXT: vrlw
; CHECK-NEXT: blr
}