caio/clang-p2996
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
d661b4b57504b965a37dc30b79bdd5ac36fca9ec
clang-p2996/llvm/test/CodeGen/AMDGPU/strict_fsub.f32.ll
Jay Foad f2c164c815 [AMDGPU] Do not wait for vscnt on function entry and return 
SIInsertWaitcnts inserts waitcnt instructions to resolve data
dependencies. The GFX10+ vscnt (VMEM store count) counter is never used
in this way. It is only used to resolve memory dependencies, and that is
handled by SIMemoryLegalizer. Hence there is no need to conservatively
wait for vscnt to be 0 on function entry and before returns.

Differential Revision: https://reviews.llvm.org/D153537
2023-07-04 12:22:38 +01:00

232 lines
10 KiB
LLVM
Raw Blame History

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -global-isel=0 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 < %s | FileCheck -check-prefix=GCN %s
; RUN: llc -global-isel=1 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 < %s | FileCheck -check-prefix=GCN %s
; RUN: llc -global-isel=0 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 < %s | FileCheck -check-prefixes=GFX10PLUS,GFX10 %s
; RUN: llc -global-isel=1 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 < %s | FileCheck -check-prefixes=GFX10PLUS,GFX10 %s
; RUN: llc -global-isel= -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -amdgpu-enable-delay-alu=0 < %s | FileCheck -check-prefixes=GFX10PLUS,GFX11 %s
; RUN: llc -global-isel=1 -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -amdgpu-enable-delay-alu=0 < %s | FileCheck -check-prefixes=GFX10PLUS,GFX11 %s
define float @v_constained_fsub_f32_fpexcept_strict(float %x, float %y) #0 {
; GCN-LABEL: v_constained_fsub_f32_fpexcept_strict:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_sub_f32_e32 v0, v0, v1
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GFX10PLUS-LABEL: v_constained_fsub_f32_fpexcept_strict:
; GFX10PLUS: ; %bb.0:
; GFX10PLUS-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10PLUS-NEXT: v_sub_f32_e32 v0, v0, v1
; GFX10PLUS-NEXT: s_setpc_b64 s[30:31]
%val = call float @llvm.experimental.constrained.fsub.f32(float %x, float %y, metadata !"round.tonearest", metadata !"fpexcept.strict")
ret float %val
}
define float @v_constained_fsub_f32_fpexcept_ignore(float %x, float %y) #0 {
; GCN-LABEL: v_constained_fsub_f32_fpexcept_ignore:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_sub_f32_e32 v0, v0, v1
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GFX10PLUS-LABEL: v_constained_fsub_f32_fpexcept_ignore:
; GFX10PLUS: ; %bb.0:
; GFX10PLUS-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10PLUS-NEXT: v_sub_f32_e32 v0, v0, v1
; GFX10PLUS-NEXT: s_setpc_b64 s[30:31]
%val = call float @llvm.experimental.constrained.fsub.f32(float %x, float %y, metadata !"round.tonearest", metadata !"fpexcept.ignore")
ret float %val
}
define float @v_constained_fsub_f32_fpexcept_maytrap(float %x, float %y) #0 {
; GCN-LABEL: v_constained_fsub_f32_fpexcept_maytrap:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_sub_f32_e32 v0, v0, v1
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GFX10PLUS-LABEL: v_constained_fsub_f32_fpexcept_maytrap:
; GFX10PLUS: ; %bb.0:
; GFX10PLUS-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10PLUS-NEXT: v_sub_f32_e32 v0, v0, v1
; GFX10PLUS-NEXT: s_setpc_b64 s[30:31]
%val = call float @llvm.experimental.constrained.fsub.f32(float %x, float %y, metadata !"round.tonearest", metadata !"fpexcept.maytrap")
ret float %val
}
define <2 x float> @v_constained_fsub_v2f32_fpexcept_strict(<2 x float> %x, <2 x float> %y) #0 {
; GCN-LABEL: v_constained_fsub_v2f32_fpexcept_strict:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_sub_f32_e32 v0, v0, v2
; GCN-NEXT: v_sub_f32_e32 v1, v1, v3
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GFX10-LABEL: v_constained_fsub_v2f32_fpexcept_strict:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10-NEXT: v_sub_f32_e32 v0, v0, v2
; GFX10-NEXT: v_sub_f32_e32 v1, v1, v3
; GFX10-NEXT: s_setpc_b64 s[30:31]
;
; GFX11-LABEL: v_constained_fsub_v2f32_fpexcept_strict:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX11-NEXT: v_dual_sub_f32 v0, v0, v2 :: v_dual_sub_f32 v1, v1, v3
; GFX11-NEXT: s_setpc_b64 s[30:31]
%val = call <2 x float> @llvm.experimental.constrained.fsub.v2f32(<2 x float> %x, <2 x float> %y, metadata !"round.tonearest", metadata !"fpexcept.strict")
ret <2 x float> %val
}
define <2 x float> @v_constained_fsub_v2f32_fpexcept_ignore(<2 x float> %x, <2 x float> %y) #0 {
; GCN-LABEL: v_constained_fsub_v2f32_fpexcept_ignore:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_sub_f32_e32 v0, v0, v2
; GCN-NEXT: v_sub_f32_e32 v1, v1, v3
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GFX10-LABEL: v_constained_fsub_v2f32_fpexcept_ignore:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10-NEXT: v_sub_f32_e32 v0, v0, v2
; GFX10-NEXT: v_sub_f32_e32 v1, v1, v3
; GFX10-NEXT: s_setpc_b64 s[30:31]
;
; GFX11-LABEL: v_constained_fsub_v2f32_fpexcept_ignore:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX11-NEXT: v_dual_sub_f32 v0, v0, v2 :: v_dual_sub_f32 v1, v1, v3
; GFX11-NEXT: s_setpc_b64 s[30:31]
%val = call <2 x float> @llvm.experimental.constrained.fsub.v2f32(<2 x float> %x, <2 x float> %y, metadata !"round.tonearest", metadata !"fpexcept.ignore")
ret <2 x float> %val
}
define <2 x float> @v_constained_fsub_v2f32_fpexcept_maytrap(<2 x float> %x, <2 x float> %y) #0 {
; GCN-LABEL: v_constained_fsub_v2f32_fpexcept_maytrap:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_sub_f32_e32 v0, v0, v2
; GCN-NEXT: v_sub_f32_e32 v1, v1, v3
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GFX10-LABEL: v_constained_fsub_v2f32_fpexcept_maytrap:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10-NEXT: v_sub_f32_e32 v0, v0, v2
; GFX10-NEXT: v_sub_f32_e32 v1, v1, v3
; GFX10-NEXT: s_setpc_b64 s[30:31]
;
; GFX11-LABEL: v_constained_fsub_v2f32_fpexcept_maytrap:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX11-NEXT: v_dual_sub_f32 v0, v0, v2 :: v_dual_sub_f32 v1, v1, v3
; GFX11-NEXT: s_setpc_b64 s[30:31]
%val = call <2 x float> @llvm.experimental.constrained.fsub.v2f32(<2 x float> %x, <2 x float> %y, metadata !"round.tonearest", metadata !"fpexcept.maytrap")
ret <2 x float> %val
}
define <3 x float> @v_constained_fsub_v3f32_fpexcept_strict(<3 x float> %x, <3 x float> %y) #0 {
; GCN-LABEL: v_constained_fsub_v3f32_fpexcept_strict:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_sub_f32_e32 v0, v0, v3
; GCN-NEXT: v_sub_f32_e32 v1, v1, v4
; GCN-NEXT: v_sub_f32_e32 v2, v2, v5
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GFX10-LABEL: v_constained_fsub_v3f32_fpexcept_strict:
; GFX10: ; %bb.0:
; GFX10-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10-NEXT: v_sub_f32_e32 v0, v0, v3
; GFX10-NEXT: v_sub_f32_e32 v1, v1, v4
; GFX10-NEXT: v_sub_f32_e32 v2, v2, v5
; GFX10-NEXT: s_setpc_b64 s[30:31]
;
; GFX11-LABEL: v_constained_fsub_v3f32_fpexcept_strict:
; GFX11: ; %bb.0:
; GFX11-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX11-NEXT: v_dual_sub_f32 v0, v0, v3 :: v_dual_sub_f32 v1, v1, v4
; GFX11-NEXT: v_sub_f32_e32 v2, v2, v5
; GFX11-NEXT: s_setpc_b64 s[30:31]
%val = call <3 x float> @llvm.experimental.constrained.fsub.v3f32(<3 x float> %x, <3 x float> %y, metadata !"round.tonearest", metadata !"fpexcept.strict")
ret <3 x float> %val
}
define amdgpu_ps float @s_constained_fsub_f32_fpexcept_strict(float inreg %x, float inreg %y) #0 {
; GCN-LABEL: s_constained_fsub_f32_fpexcept_strict:
; GCN: ; %bb.0:
; GCN-NEXT: v_mov_b32_e32 v0, s3
; GCN-NEXT: v_sub_f32_e32 v0, s2, v0
; GCN-NEXT: ; return to shader part epilog
;
; GFX10PLUS-LABEL: s_constained_fsub_f32_fpexcept_strict:
; GFX10PLUS: ; %bb.0:
; GFX10PLUS-NEXT: v_sub_f32_e64 v0, s2, s3
; GFX10PLUS-NEXT: ; return to shader part epilog
%val = call float @llvm.experimental.constrained.fsub.f32(float %x, float %y, metadata !"round.tonearest", metadata !"fpexcept.strict")
ret float %val
}
define float @v_constained_fsub_f32_fpexcept_strict_fabs_lhs(float %x, float %y) #0 {
; GCN-LABEL: v_constained_fsub_f32_fpexcept_strict_fabs_lhs:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_sub_f32_e64 v0, |v0|, v1
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GFX10PLUS-LABEL: v_constained_fsub_f32_fpexcept_strict_fabs_lhs:
; GFX10PLUS: ; %bb.0:
; GFX10PLUS-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10PLUS-NEXT: v_sub_f32_e64 v0, |v0|, v1
; GFX10PLUS-NEXT: s_setpc_b64 s[30:31]
%fabs.x = call float @llvm.fabs.f32(float %x)
%val = call float @llvm.experimental.constrained.fsub.f32(float %fabs.x, float %y, metadata !"round.tonearest", metadata !"fpexcept.strict")
ret float %val
}
define float @v_constained_fsub_f32_fpexcept_strict_fabs_rhs(float %x, float %y) #0 {
; GCN-LABEL: v_constained_fsub_f32_fpexcept_strict_fabs_rhs:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_sub_f32_e64 v0, v0, |v1|
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GFX10PLUS-LABEL: v_constained_fsub_f32_fpexcept_strict_fabs_rhs:
; GFX10PLUS: ; %bb.0:
; GFX10PLUS-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10PLUS-NEXT: v_sub_f32_e64 v0, v0, |v1|
; GFX10PLUS-NEXT: s_setpc_b64 s[30:31]
%fabs.y = call float @llvm.fabs.f32(float %y)
%val = call float @llvm.experimental.constrained.fsub.f32(float %x, float %fabs.y, metadata !"round.tonearest", metadata !"fpexcept.strict")
ret float %val
}
define float @v_constained_fsub_f32_fpexcept_strict_fneg_fabs_lhs(float %x, float %y) #0 {
; GCN-LABEL: v_constained_fsub_f32_fpexcept_strict_fneg_fabs_lhs:
; GCN: ; %bb.0:
; GCN-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GCN-NEXT: v_sub_f32_e64 v0, -|v0|, v1
; GCN-NEXT: s_setpc_b64 s[30:31]
;
; GFX10PLUS-LABEL: v_constained_fsub_f32_fpexcept_strict_fneg_fabs_lhs:
; GFX10PLUS: ; %bb.0:
; GFX10PLUS-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX10PLUS-NEXT: v_sub_f32_e64 v0, -|v0|, v1
; GFX10PLUS-NEXT: s_setpc_b64 s[30:31]
%fabs.x = call float @llvm.fabs.f32(float %x)
%neg.fabs.x = fneg float %fabs.x
%val = call float @llvm.experimental.constrained.fsub.f32(float %neg.fabs.x, float %y, metadata !"round.tonearest", metadata !"fpexcept.strict")
ret float %val
}
declare float @llvm.fabs.f32(float) #1
declare float @llvm.experimental.constrained.fsub.f32(float, float, metadata, metadata) #1
declare <2 x float> @llvm.experimental.constrained.fsub.v2f32(<2 x float>, <2 x float>, metadata, metadata) #1
declare <3 x float> @llvm.experimental.constrained.fsub.v3f32(<3 x float>, <3 x float>, metadata, metadata) #1
attributes #0 = { strictfp }
attributes #1 = { inaccessiblememonly nounwind willreturn }
Reference in New Issue View Git Blame Copy Permalink