Skip to content

[X86] Do not directly fold for VINSERTPS #65718

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
phoebewang merged 2 commits into from
Sep 8, 2023
Merged

[X86] Do not directly fold for VINSERTPS #65718

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
cb3682e
[X86] Do not directly fold for VINSERTPS
phoebewang Sep 8, 2023
4d5bf91
Move to INSERTPSrr together
phoebewang Sep 8, 2023
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/test/CodeGen/X86/avx.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -184,12 +184,14 @@ define <4 x float> @nofold_insertps(ptr %a, <4 x float> %b) {
; X86-LABEL: nofold_insertps:
; X86: ## %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: vinsertps $176, (%eax), %xmm0, %xmm0 ## xmm0 = xmm0[0,1,2],mem[2]
; X86-NEXT: vmovups (%eax), %xmm1
; X86-NEXT: vinsertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[2]
; X86-NEXT: retl
;
; X64-LABEL: nofold_insertps:
; X64: ## %bb.0:
; X64-NEXT: vinsertps $176, (%rdi), %xmm0, %xmm0 ## xmm0 = xmm0[0,1,2],mem[2]
; X64-NEXT: vmovups (%rdi), %xmm1
; X64-NEXT: vinsertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[2]
; X64-NEXT: retq
%1 = load <4 x float>, ptr %a, align 1
%2 = shufflevector <4 x float> %b, <4 x float> %1, <4 x i32> <i32 0, i32 1, i32 2, i32 6>
Expand Down
2 changes: 0 additions & 2 deletions llvm/test/TableGen/x86-fold-tables.inc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2105,8 +2105,6 @@ static const X86MemoryFoldTableEntry MemoryFoldTable2[] = {
{X86::VINSERTI64x2Z256rr, X86::VINSERTI64x2Z256rm, 0},
{X86::VINSERTI64x2Zrr, X86::VINSERTI64x2Zrm, 0},
{X86::VINSERTI64x4Zrr, X86::VINSERTI64x4Zrm, 0},
{X86::VINSERTPSZrr, X86::VINSERTPSZrm, TB_NO_REVERSE},
{X86::VINSERTPSrr, X86::VINSERTPSrm, TB_NO_REVERSE},
{X86::VMAXCPDYrr, X86::VMAXCPDYrm, 0},
{X86::VMAXCPDZ128rr, X86::VMAXCPDZ128rm, 0},
{X86::VMAXCPDZ256rr, X86::VMAXCPDZ256rm, 0},
Expand Down
2 changes: 2 additions & 0 deletions llvm/utils/TableGen/X86ManualFoldTables.def
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,6 +43,8 @@ NOFOLD(VEXTRACTI32x8Zrrk)
NOFOLD(VEXTRACTI64x2Z256rrk)
NOFOLD(VEXTRACTI64x2Zrrk)
NOFOLD(VEXTRACTI64x4Zrrk)
NOFOLD(VINSERTPSZrr)
Copy link
Contributor

@KanRobert KanRobert Sep 8, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could you move this to line 239 NOFOLD(INSERTPSrr)? So we share the comments

Copy link
Contributor Author

@phoebewang phoebewang Sep 8, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good point, I didn't notice it before

NOFOLD(VINSERTPSrr)
NOFOLD(VMOVAPDZ128mrk)
NOFOLD(VMOVAPDZ256mrk)
NOFOLD(VMOVAPDZmrk)
Expand Down