[REVIEW] Use CMAKE_CUDA_ARCHITECTURES - #7391
Conversation
github-actions
Bot
added
CMake
kkraus14
left a comment
kkraus14
left a comment
There was a problem hiding this comment.
Things generally look good. Should we consider to move to using the CUDA_ARCHITECTURES target property instead of manually passing flags to the CUDA compiler?
I think that move is reasonable. Maybe after / part of #7107? It would allow better compilation flags control. If we ever needed targets built for a subset of archs we could. |
Codecov Report
@@ Coverage Diff @@
## branch-0.19 #7391 +/- ##
==============================================
Coverage ? 82.19%
==============================================
Files ? 100
Lines ? 16968
Branches ? 0
==============================================
Hits ? 13947
Misses ? 3021
Partials ? 0 Continue to review full report at Codecov.
|
harrism
added
non-breaking
harrism
changed the title
|
rerun tests |
robertmaynard
changed the title
473169d to
db5c7d3
Compare
db5c7d3 to
d319394
Compare
robertmaynard
changed the title
kkraus14
added
breaking
|
Pinged some folks offline regarding rapids-compose and devops scripts to make sure no one is using |
|
@gpucibot merge |
Based on #706 as they both modify `cmake/Modules/SetGPUArchs.cmake` This brings rmm's handling of `CMAKE_CUDA_ARCHITECTURES` to match that is proposed for cudf in NVIDIA/cudf#7391 Authors: - Robert Maynard (@robertmaynard) Approvers: - Mark Harris (@harrism) - Keith Kraus (@kkraus14) URL: #709
This eliminates the `Policy CMP0104 is not set` warning during the JNI build by using `CMAKE_CUDA_ARCHITECTURES` to specify the targeted CUDA architectures during the build. This eliminates a lot of code replicated from the cpp build and reuses the `ConfigureCUDA` module added in #7391. The architectures being used by the build are visible in the `mvn` output, e.g.: ``` [exec] -- CUDF: Building CUDF for GPU architectures: 60-real;70-real;75 ``` This also configures the CPU compiler to use the same flags as the cpp build which required fixing a number of warnings (e.g.: sign mismatch comparisons, unused variables, etc.) in the JNI code since warnings are errors in the build. Authors: - Jason Lowe (@jlowe) Approvers: - Robert (Bobby) Evans (@revans2) - Alessandro Bellina (@abellina) - MithunR (@mythrocks) URL: #7425
4 participants
When using CMake >= 3.18 you now don't get
CMP0104warnings for each target that use CUDA.To correct this issue I have cherry-picked a collection of changes from #7107 which updated cudf to use
CMAKE_CUDA_ARCHITECTURES.When using CMake < 3.18 cudf will transform
CMAKE_CUDA_ARCHITECTURESinto the correct flags and store it intoCMAKE_CUDA_FLAGS.When I was porting this over from !7107 I noticed that when cudf is built for multi arch ( sm60, sm70, sm80, ... ) it only compiles compute for the newest arch. This behavior is preserved, and !7107 will need to be updated with this change.