Rebuild ESPResSo for CUDA sanity check#1168
Merged
bedroge merged 4 commits intoEESSI:mainfrom Sep 18, 2025
Merged
Conversation
Collaborator
Author
|
bot: build repo:eessi.io-2023.06-software instance:eessi-bot-surf for:arch=x86_64/intel/icelake,accel=nvidia/cc80 |
|
New job on instance
|
Collaborator
Author
|
Collaborator
Author
Hm... seems to be compiled for SM_52, and with PTX code for SM_52. So, something needs to be fixed here: how do we convince the ESPResSo build system to compile for a different CUDA arch? |
Collaborator
|
Seems like that's hardcoded here for version 4.2.2: Though new versions apparently will allow you to explicitly define this: |
Contributor
|
The following patch should work: diff --git a/cmake/FindCUDACompilerNVCC.cmake b/cmake/FindCUDACompilerNVCC.cmake
index 08f9db312..f68d4db94 100644
--- a/cmake/FindCUDACompilerNVCC.cmake
+++ b/cmake/FindCUDACompilerNVCC.cmake
@@ -52,8 +52,16 @@ list(APPEND CUDA_NVCC_FLAGS_COVERAGE -O3 -g -Xptxas=-O3 -Xcompiler=-Og,-g)
list(APPEND CUDA_NVCC_FLAGS_RELWITHASSERT -O3 -g -Xptxas=-O3 -Xcompiler=-O3,-g)
-if(CMAKE_CUDA_COMPILER_VERSION VERSION_LESS 11)
- list(APPEND CUDA_NVCC_FLAGS -gencode=arch=compute_30,code=sm_30)
+if(NOT DEFINED ESPRESSO_CUDA_ARCHITECTURES)
+ if("$ENV{CUDAARCHS}" STREQUAL "")
+ set(ESPRESSO_CUDA_ARCHITECTURES "75;86;89" CACHE INTERNAL "")
+ else()
+ set(ESPRESSO_CUDA_ARCHITECTURES "$ENV{CUDAARCHS}" CACHE INTERNAL "")
+ endif()
endif()
+foreach(ESPRESSO_CUDA_ARCH ${ESPRESSO_CUDA_ARCHITECTURES})
+ list(APPEND CUDA_NVCC_FLAGS
+ "-gencode=arch=compute_${ESPRESSO_CUDA_ARCH},code=sm_${ESPRESSO_CUDA_ARCH}"
+ "-gencode=arch=compute_${ESPRESSO_CUDA_ARCH},code=compute_${ESPRESSO_CUDA_ARCH}")
+endforeach()
list(APPEND CUDA_NVCC_FLAGS
- -gencode=arch=compute_52,code=sm_52
- -gencode=arch=compute_52,code=compute_52 -std=c++${CMAKE_CUDA_STANDARD}
+ -std=c++${CMAKE_CUDA_STANDARD}
$<$<BOOL:${WARNINGS_ARE_ERRORS}>:-Xcompiler=-Werror;-Xptxas=-Werror>With this one can invoke the |
bedroge
reviewed
Sep 5, 2025
…50828-eb-5.1.1-rebuild-ESPResSo-for-cuda-sanity-check.yml
Collaborator
|
bot: build repo:eessi.io-2023.06-software instance:eessi-bot-surf for:arch=x86_64/intel/icelake,accel=nvidia/cc80 |
|
New job on instance
|
bedroge
reviewed
Sep 5, 2025
Collaborator
|
bot: build repo:eessi.io-2023.06-software instance:eessi-bot-surf for:arch=x86_64/intel/icelake,accel=nvidia/cc80 |
|
New job on instance
|
Collaborator
edit: oh, of course, we need to set |
Contributor
|
Or pass |
bedroge
reviewed
Sep 5, 2025
Collaborator
|
bot: build repo:eessi.io-2023.06-software instance:eessi-bot-surf for:arch=x86_64/intel/icelake,accel=nvidia/cc80 |
|
New job on instance
|
Collaborator
|
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
Collaborator
|
bot: build repo:eessi.io-2023.06-software instance:eessi-bot-mc-aws on:arch=zen2 for:arch=x86_64/amd/zen2,accel=nvidia/cc70 |
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
Merged
Collaborator
|
Only need to do one more build with the Surf bot, but the cluster is currently down. |
Collaborator
|
Snellius is back online. Not sure if the bot is already available, but let's give it a try: bot: build repo:eessi.io-2023.06-software instance:eessi-bot-surf for:arch=x86_64/amd/zen4,accel=nvidia/cc90 |
|
New job on instance
|
Collaborator
|
The Surf job was OOM killed. Don't really understand this, the other job that ran on Snellius (for icelake+cc80, see #1168 (comment)) only used: |
Collaborator
|
Let's try it one more time: bot: build repo:eessi.io-2023.06-software instance:eessi-bot-surf for:arch=x86_64/amd/zen4,accel=nvidia/cc90 |
|
New job on instance
|
bedroge
added
bot:deploy
Collaborator
|
Staging PR merged, all tarballs have been ingested 🎉 |
bedroge
approved these changes
Sep 18, 2025
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Note: I've seen in an interactive build that this didn't pass the CUDA sanity check. So we may have to investigate why not (i.e. which files don't provide the correct device code, and why), and how to make it pass...