Skip to content

[AMDGPU] Add VOPD combine dependency tests. NFC. #104841

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.

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 1 commit into from
Aug 19, 2024
Merged
Changes from all commits
Commits
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
49 changes: 49 additions & 0 deletions llvm/test/CodeGen/AMDGPU/vopd-combine.mir
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,8 @@
define void @vopd_mov_fixup_fail() { ret void }
define void @vopd_no_combine_dependent_subreg() { ret void }
define void @vopd_mov_mov_same_src_bank() { ret void }
define void @vopd_combine_opy_overwrites_opx() { ret void }
define void @vopd_no_combine_opx_overwrites_opy() { ret void }
...

---
Expand Down Expand Up @@ -737,3 +739,50 @@ body: |
$vgpr2 = V_MOV_B32_e32 $vgpr1, implicit $exec
$vgpr3 = V_MOV_B32_e32 $vgpr5, implicit $exec
...

---
# There is no race between mov and add, they are executed at the same time
name: vopd_combine_opy_overwrites_opx
tracksRegLiveness: true
body: |
bb.0:
; SCHED-LABEL: name: vopd_combine_opy_overwrites_opx
; SCHED: $vgpr1 = IMPLICIT_DEF
; SCHED-NEXT: $vgpr3 = IMPLICIT_DEF
; SCHED-NEXT: $vgpr0 = V_MOV_B32_e32 killed $vgpr1, implicit $exec
; SCHED-NEXT: $vgpr1 = V_ADD_F32_e32 killed $vgpr3, $vgpr3, implicit $mode, implicit $exec
;
; PAIR-GFX11-LABEL: name: vopd_combine_opy_overwrites_opx
; PAIR-GFX11: $vgpr1 = IMPLICIT_DEF
; PAIR-GFX11-NEXT: $vgpr3 = IMPLICIT_DEF
; PAIR-GFX11-NEXT: $vgpr0, $vgpr1 = V_DUAL_MOV_B32_e32_X_ADD_F32_e32_gfx11 killed $vgpr1, killed $vgpr3, $vgpr3, implicit $exec, implicit $mode, implicit $exec, implicit $mode, implicit $exec
;
; PAIR-GFX12-LABEL: name: vopd_combine_opy_overwrites_opx
; PAIR-GFX12: $vgpr1 = IMPLICIT_DEF
; PAIR-GFX12-NEXT: $vgpr3 = IMPLICIT_DEF
; PAIR-GFX12-NEXT: $vgpr0, $vgpr1 = V_DUAL_MOV_B32_e32_X_ADD_F32_e32_gfx12 killed $vgpr1, killed $vgpr3, $vgpr3, implicit $exec, implicit $mode, implicit $exec, implicit $mode, implicit $exec
$vgpr1 = IMPLICIT_DEF
$vgpr3 = IMPLICIT_DEF
$vgpr0 = V_MOV_B32_e32 $vgpr1, implicit $exec
$vgpr1 = V_ADD_F32_e32 $vgpr3, $vgpr3, implicit $mode, implicit $exec
...

---
# OPY cannot cosume value produced by OPX
name: vopd_no_combine_opx_overwrites_opy
tracksRegLiveness: true
body: |
bb.0:
; SCHED-LABEL: name: vopd_no_combine_opx_overwrites_opy
; SCHED: $vgpr1 = IMPLICIT_DEF
; SCHED-NEXT: $vgpr0 = V_MOV_B32_e32 $vgpr1, implicit $exec
; SCHED-NEXT: $vgpr3 = V_ADD_F32_e32 killed $vgpr0, killed $vgpr1, implicit $mode, implicit $exec
;
; PAIR-LABEL: name: vopd_no_combine_opx_overwrites_opy
; PAIR: $vgpr1 = IMPLICIT_DEF
; PAIR-NEXT: $vgpr0 = V_MOV_B32_e32 $vgpr1, implicit $exec
; PAIR-NEXT: $vgpr3 = V_ADD_F32_e32 killed $vgpr0, killed $vgpr1, implicit $mode, implicit $exec
$vgpr1 = IMPLICIT_DEF
$vgpr0 = V_MOV_B32_e32 $vgpr1, implicit $exec
$vgpr3 = V_ADD_F32_e32 $vgpr0, $vgpr1, implicit $mode, implicit $exec
...
Loading