[llvm] r174330 - Disable a couple more vector splat optimizations on PPC.
Benjamin Kramer
benny.kra at googlemail.com
Mon Feb 4 07:52:32 PST 2013
Author: d0k
Date: Mon Feb 4 09:52:32 2013
New Revision: 174330
URL: http://llvm.org/viewvc/llvm-project?rev=174330&view=rev
Log:
Disable a couple more vector splat optimizations on PPC.
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.
Modified:
llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp
llvm/trunk/test/CodeGen/PowerPC/vec_constants.ll
Modified: llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp?rev=174330&r1=174329&r2=174330&view=diff
==============================================================================
--- llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/PowerPC/PPCISelLowering.cpp Mon Feb 4 09:52:32 2013
@@ -5138,14 +5138,16 @@ SDValue PPCTargetLowering::LowerBUILD_VE
// 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);
Modified: llvm/trunk/test/CodeGen/PowerPC/vec_constants.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/vec_constants.ll?rev=174330&r1=174329&r2=174330&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/PowerPC/vec_constants.ll (original)
+++ llvm/trunk/test/CodeGen/PowerPC/vec_constants.ll Mon Feb 4 09:52:32 2013
@@ -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
%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
}
More information about the llvm-commits
mailing list