replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[InstCombine] Turn (extractelement <1 x i64/double> (bitcast (x86_mmx))) into a single bitcast from x86_mmx to i64/double. 

The _m64 type is represented in IR as <1 x i64>. The x86-64 ABI
on Linux passes <1 x i64> as a double. MMX intrinsics use x86_mmx
type in IR.These things result in a lot of bitcasts in mmx code.
There's another instcombine that tries to turn bitcast <1 x i64>
to double into extractelement and a bitcast.

The combine here tries to reverse this extractelement conversion
if we see an mmx type.
This commit is contained in:
Craig Topper 2019-11-10 16:20:19 -08:00
parent d37db750c2
commit aafde063aa
2 changed files with 8 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
llvm/lib/Transforms/InstCombine/InstCombineVectorOps.cpp
View File

@ -435,6 +435,13 @@ Instruction *InstCombiner::visitExtractElementInst(ExtractElementInst &EI) {
Worklist.AddValue(EE);
return CastInst::Create(CI->getOpcode(), EE, EI.getType());
}
// If the input is a bitcast from x86_mmx, turn into a single bitcast from
// the mmx type to the scalar type.
if (CI->getOpcode() == Instruction::BitCast &&
EI.getVectorOperandType()->getNumElements() == 1 &&
CI->getOperand(0)->getType()->isX86_MMXTy())
return new BitCastInst(CI->getOperand(0), EI.getType());
}
}
return nullptr;

3
llvm/test/Transforms/InstCombine/bitcast-vec-canon.ll
View File

@ -59,8 +59,7 @@ define <1 x i64> @f(x86_mmx %y) {
define double @g(x86_mmx %x) {
; CHECK-LABEL: @g(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[BC:%.*]] = bitcast x86_mmx %x to <1 x double>
; CHECK-NEXT: [[TMP0:%.*]] = extractelement <1 x double> [[BC]], i32 0
; CHECK-NEXT: [[TMP0:%.*]] = bitcast x86_mmx [[X:%.*]] to double
; CHECK-NEXT: ret double [[TMP0]]
;
entry: