This is an archive of the discontinued LLVM Phabricator instance.

Differential D156936

[Clang] Increase default architecture from sm_35 to sm_52
ClosedPublic

Authored by jhuber6 on Aug 2 2023, 1:32 PM.

Details

Reviewers
tra
jdoerfert
yaxunl
jlebar
Commits
rGab202aa7004a: [Clang] Increase default architecture from sm_35 to sm_52
Summary

We previously defaulted to sm_35 for the purpose of unspecified
architecture. This was removed in new CUDA versions so we should bump
this up.

Diff Detail

Event Timeline

jhuber6 created this revision.Aug 2 2023, 1:32 PM
Herald added a project: Restricted Project. · View Herald TranscriptAug 2 2023, 1:32 PM
jhuber6 requested review of this revision.Aug 2 2023, 1:32 PM
Herald added a project: Restricted Project. · View Herald TranscriptAug 2 2023, 1:32 PM
Herald added a subscriber: cfe-commits. · View Herald Transcript
jdoerfert accepted this revision.Aug 2 2023, 2:45 PM
This revision is now accepted and ready to land.Aug 2 2023, 2:45 PM
This revision was landed with ongoing or failed builds.Aug 2 2023, 3:30 PM
Closed by commit rGab202aa7004a: [Clang] Increase default architecture from sm_35 to sm_52 (authored by jhuber6). · Explain Why
This revision was automatically updated to reflect the committed changes.
jhuber6 added a commit: rGab202aa7004a: [Clang] Increase default architecture from sm_35 to sm_52.
Harbormaster completed remote builds in B249869: Diff 546589.Aug 2 2023, 5:02 PM

Revision Contents

PathSize
clang/
include/
clang/
Basic/
2 lines
test/
Driver/
6 lines

Diff 546622

clang/include/clang/Basic/Cuda.h

Show First 20 Lines • Show All 111 Lines • ▼ Show 20 Linesenum class CudaArch {
GFX1102, GFX1102,
GFX1103, GFX1103,
GFX1150, GFX1150,
GFX1151, GFX1151,
Generic, // A processor model named 'generic' if the target backend defines a Generic, // A processor model named 'generic' if the target backend defines a
// public one. // public one.
LAST, LAST,
CudaDefault = CudaArch::SM_35, CudaDefault = CudaArch::SM_52,
HIPDefault = CudaArch::GFX803, HIPDefault = CudaArch::GFX803,
}; };
static inline bool IsNVIDIAGpuArch(CudaArch A) { static inline bool IsNVIDIAGpuArch(CudaArch A) {
return A >= CudaArch::SM_20 && A < CudaArch::GFX600; return A >= CudaArch::SM_20 && A < CudaArch::GFX600;
} }
static inline bool IsAMDGpuArch(CudaArch A) { static inline bool IsAMDGpuArch(CudaArch A) {
Show All 31 Lines

clang/test/Driver/cuda-cross-compiling.c

Show First 20 Lines • Show All 59 Lines • ▼ Show 20 Lines
// LINK: nvlink{{.*}}"-o" "a.out" "-arch" "sm_61" {{.*}} "{{.*}}.cubin" // LINK: nvlink{{.*}}"-o" "a.out" "-arch" "sm_61" {{.*}} "{{.*}}.cubin"
// //
// Test the generated arguments default to a value with no architecture. // Test the generated arguments default to a value with no architecture.
// //
// RUN: %clang --target=nvptx64-nvidia-cuda -### --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \ // RUN: %clang --target=nvptx64-nvidia-cuda -### --cuda-path=%S/Inputs/CUDA/usr/local/cuda %s 2>&1 \
// RUN: | FileCheck -check-prefix=DEFAULT %s // RUN: | FileCheck -check-prefix=DEFAULT %s
// DEFAULT: -cc1" "-triple" "nvptx64-nvidia-cuda" "-S" {{.*}} "-target-cpu" "sm_35" "-target-feature" "+ptx{{[0-9]+}}" {{.*}} "-o" "[[PTX:.+]].s" // DEFAULT: -cc1" "-triple" "nvptx64-nvidia-cuda" "-S" {{.*}} "-target-cpu" "sm_52" "-target-feature" "+ptx{{[0-9]+}}" {{.*}} "-o" "[[PTX:.+]].s"
// DEFAULT-NEXT: ptxas{{.*}}"-m64" "-O0" "--gpu-name" "sm_35" "--output-file" "[[CUBIN:.+]].cubin" "[[PTX]].s" "-c" // DEFAULT-NEXT: ptxas{{.*}}"-m64" "-O0" "--gpu-name" "sm_52" "--output-file" "[[CUBIN:.+]].cubin" "[[PTX]].s" "-c"
// DEFAULT-NEXT: nvlink{{.*}}"-o" "a.out" "-arch" "sm_35" {{.*}} "[[CUBIN]].cubin" // DEFAULT-NEXT: nvlink{{.*}}"-o" "a.out" "-arch" "sm_52" {{.*}} "[[CUBIN]].cubin"
// //
// Test to ensure that we enable handling global constructors in a freestanding // Test to ensure that we enable handling global constructors in a freestanding
// Nvidia compilation. // Nvidia compilation.
// //
// RUN: %clang -target nvptx64-nvidia-cuda -march=sm_70 %s -### 2>&1 \ // RUN: %clang -target nvptx64-nvidia-cuda -march=sm_70 %s -### 2>&1 \
// RUN: | FileCheck -check-prefix=LOWERING %s // RUN: | FileCheck -check-prefix=LOWERING %s
// LOWERING: -cc1" "-triple" "nvptx64-nvidia-cuda" {{.*}} "-mllvm" "--nvptx-lower-global-ctor-dtor" // LOWERING: -cc1" "-triple" "nvptx64-nvidia-cuda" {{.*}} "-mllvm" "--nvptx-lower-global-ctor-dtor"