This is an archive of the discontinued LLVM Phabricator instance.

Differential D74400

AMDGPU: llvm.amdgcn.writelane is a source of divergence
ClosedPublic

Authored by nhaehnle on Feb 11 2020, 5:42 AM.

Details

Reviewers
arsenm
foad
mareko
Commits
rGab2f610f3855: AMDGPU: llvm.amdgcn.writelane is a source of divergence
Summary

Consider:

%r = call i32 @llvm.amdgcn.writelane(i32 0, i32 1, i32 2)

This produces a value that is 0 on lane 1, and 2 everywhere else; i.e.,
it is divergent.

Reported-by: Marek Olsak <Marek.Olsak@amd.com>

Diff Detail

Unit TestsFailed

TimeTest
2,190 mslibc++.std/thread/thread_mutex/thread_mutex_requirements/thread_timedmutex_requirements/thread_timedmutex_recursive::Unknown Unit Message ("")

Event Timeline

nhaehnle created this revision.Feb 11 2020, 5:42 AM
Herald added a project: Restricted Project. · View Herald TranscriptFeb 11 2020, 5:42 AM
Herald added subscribers: kerbowa, hiraditya, t-tye and 6 others. · View Herald Transcript
Harbormaster failed remote builds in B46216: Diff 243836!Feb 11 2020, 6:01 AM
foad accepted this revision.Feb 11 2020, 6:04 AM

LGTM.

This revision is now accepted and ready to land.Feb 11 2020, 6:04 AM
Closed by commit rGab2f610f3855: AMDGPU: llvm.amdgcn.writelane is a source of divergence (authored by nhaehnle). · Explain WhyFeb 12 2020, 12:16 AM
This revision was automatically updated to reflect the committed changes.

Revision Contents

PathSize
llvm/
lib/
Target/
AMDGPU/
1 line
test/
Analysis/
DivergenceAnalysis/
AMDGPU/
8 lines

Diff 243836

llvm/lib/Target/AMDGPU/AMDGPUSearchableTables.td

Show First 20 Lines • Show All 241 Lines • ▼ Show 20 Lines
def : SourceOfDivergence<int_amdgcn_ds_swizzle>; def : SourceOfDivergence<int_amdgcn_ds_swizzle>;
def : SourceOfDivergence<int_amdgcn_ds_ordered_add>; def : SourceOfDivergence<int_amdgcn_ds_ordered_add>;
def : SourceOfDivergence<int_amdgcn_ds_ordered_swap>; def : SourceOfDivergence<int_amdgcn_ds_ordered_swap>;
def : SourceOfDivergence<int_amdgcn_permlane16>; def : SourceOfDivergence<int_amdgcn_permlane16>;
def : SourceOfDivergence<int_amdgcn_permlanex16>; def : SourceOfDivergence<int_amdgcn_permlanex16>;
def : SourceOfDivergence<int_amdgcn_mov_dpp>; def : SourceOfDivergence<int_amdgcn_mov_dpp>;
def : SourceOfDivergence<int_amdgcn_mov_dpp8>; def : SourceOfDivergence<int_amdgcn_mov_dpp8>;
def : SourceOfDivergence<int_amdgcn_update_dpp>; def : SourceOfDivergence<int_amdgcn_update_dpp>;
def : SourceOfDivergence<int_amdgcn_writelane>;
def : SourceOfDivergence<int_amdgcn_mfma_f32_4x4x1f32>; def : SourceOfDivergence<int_amdgcn_mfma_f32_4x4x1f32>;
def : SourceOfDivergence<int_amdgcn_mfma_f32_4x4x1f32>; def : SourceOfDivergence<int_amdgcn_mfma_f32_4x4x1f32>;
def : SourceOfDivergence<int_amdgcn_mfma_f32_4x4x4f16>; def : SourceOfDivergence<int_amdgcn_mfma_f32_4x4x4f16>;
def : SourceOfDivergence<int_amdgcn_mfma_i32_4x4x4i8>; def : SourceOfDivergence<int_amdgcn_mfma_i32_4x4x4i8>;
def : SourceOfDivergence<int_amdgcn_mfma_f32_4x4x2bf16>; def : SourceOfDivergence<int_amdgcn_mfma_f32_4x4x2bf16>;
def : SourceOfDivergence<int_amdgcn_mfma_f32_16x16x1f32>; def : SourceOfDivergence<int_amdgcn_mfma_f32_16x16x1f32>;
def : SourceOfDivergence<int_amdgcn_mfma_f32_16x16x4f32>; def : SourceOfDivergence<int_amdgcn_mfma_f32_16x16x4f32>;
Show All 25 Lines

llvm/test/Analysis/DivergenceAnalysis/AMDGPU/intrinsics.ll

Show All 36 Lines
; CHECK: DIVERGENT: %tmp0 = call i32 @llvm.amdgcn.mov.dpp8.i32(i32 %in, i32 1) #0 ; CHECK: DIVERGENT: %tmp0 = call i32 @llvm.amdgcn.mov.dpp8.i32(i32 %in, i32 1) #0
define amdgpu_kernel void @mov_dpp8(i32 addrspace(1)* %out, i32 %in) #0 { define amdgpu_kernel void @mov_dpp8(i32 addrspace(1)* %out, i32 %in) #0 {
%tmp0 = call i32 @llvm.amdgcn.mov.dpp8.i32(i32 %in, i32 1) #0 %tmp0 = call i32 @llvm.amdgcn.mov.dpp8.i32(i32 %in, i32 1) #0
store i32 %tmp0, i32 addrspace(1)* %out store i32 %tmp0, i32 addrspace(1)* %out
ret void ret void
} }
; CHECK: DIVERGENT: %tmp0 = call i32 @llvm.amdgcn.writelane(i32 0, i32 1, i32 2)
define amdgpu_kernel void @writelane(i32 addrspace(1)* %out) #0 {
%tmp0 = call i32 @llvm.amdgcn.writelane(i32 0, i32 1, i32 2)
store i32 %tmp0, i32 addrspace(1)* %out
ret void
}
declare i32 @llvm.amdgcn.ds.swizzle(i32, i32) #1 declare i32 @llvm.amdgcn.ds.swizzle(i32, i32) #1
declare i32 @llvm.amdgcn.permlane16(i32, i32, i32, i32, i1, i1) #1 declare i32 @llvm.amdgcn.permlane16(i32, i32, i32, i32, i1, i1) #1
declare i32 @llvm.amdgcn.permlanex16(i32, i32, i32, i32, i1, i1) #1 declare i32 @llvm.amdgcn.permlanex16(i32, i32, i32, i32, i1, i1) #1
declare i32 @llvm.amdgcn.mov.dpp.i32(i32, i32, i32, i32, i1) #1 declare i32 @llvm.amdgcn.mov.dpp.i32(i32, i32, i32, i32, i1) #1
declare i32 @llvm.amdgcn.mov.dpp8.i32(i32, i32) #1 declare i32 @llvm.amdgcn.mov.dpp8.i32(i32, i32) #1
declare i32 @llvm.amdgcn.update.dpp.i32(i32, i32, i32, i32, i32, i1) #1 declare i32 @llvm.amdgcn.update.dpp.i32(i32, i32, i32, i32, i32, i1) #1
declare i32 @llvm.amdgcn.writelane(i32, i32, i32) #1
attributes #0 = { nounwind convergent } attributes #0 = { nounwind convergent }
attributes #1 = { nounwind readnone convergent } attributes #1 = { nounwind readnone convergent }