Skip to content

[CodeGen] Fix generating permute bytes from register pair when the initial values are undefined #74437

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jan 16, 2024
Merged

[CodeGen] Fix generating permute bytes from register pair when the initial values are undefined #74437

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
27afcf4
Fix generating permute bytes from register pair when the initial valu…
mmoadeli Dec 5, 2023
3452d87
Add comment to clarify -1 as a placeholder for undef value.
mmoadeli Jan 15, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 4 additions & 2 deletions llvm/lib/Target/NVPTX/NVPTXISelLowering.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2360,8 +2360,10 @@ SDValue NVPTXTargetLowering::LowerVECTOR_SHUFFLE(SDValue Op,
const ShuffleVectorSDNode *SVN = cast<ShuffleVectorSDNode>(Op.getNode());
SDValue V2 = Op.getOperand(1);
uint32_t Selector = 0;
for (auto I : llvm::enumerate(SVN->getMask()))
Selector |= (I.value() << (I.index() * 4));
for (auto I : llvm::enumerate(SVN->getMask())) {
if (I.value() != -1) // -1 is a placeholder for undef.
Selector |= (I.value() << (I.index() * 4));
}

SDLoc DL(Op);
return DAG.getNode(NVPTXISD::PRMT, DL, MVT::v4i8, V1, V2,
Expand Down
18 changes: 18 additions & 0 deletions llvm/test/CodeGen/NVPTX/shuffle-vec-undef-init.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,18 @@
; RUN: llc < %s -march=nvptx -mcpu=sm_20 -verify-machineinstrs | FileCheck %s
; RUN: llc < %s -march=nvptx -mcpu=sm_20 -verify-machineinstrs | FileCheck %s -check-prefix=CHECK-FOUND

define void @kernel_func(ptr %in.vec, ptr %out.vec0) nounwind {
entry:
%wide.vec = load <32 x i8>, ptr %in.vec, align 64
%vec0 = shufflevector <32 x i8> %wide.vec, <32 x i8> undef, <4 x i32> <i32 0, i32 8, i32 16, i32 24>
store <4 x i8> %vec0, ptr %out.vec0, align 64
ret void

; CHECK-FOUND: prmt.b32 {{.*}} 16384;
; CHECK-FOUND: prmt.b32 {{.*}} 64;
; CHECK-FOUND: prmt.b32 {{.*}} 30224;

; CHECK: @kernel_func
; CHECK-NOT: prmt.b32 {{.*}} -1;
; CHECK: -- End function
}