New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Revert to CUDA 10.1 Update 1 (10.1.168), add support for Power (11.0.x backport) #5419
Revert to CUDA 10.1 Update 1 (10.1.168), add support for Power (11.0.x backport) #5419
Conversation
A new Pull Request was created by @fwyzard (Andrea Bocci) for branch IB/CMSSW_11_0_X/master. @cmsbuild, @smuzaffar, @mrodozov can you please review it and eventually sign? Thanks. |
please test |
The tests are being triggered in jenkins. |
using following comment (as mentioned http://cms-sw.github.io/cms-bot-cmssw-cmds.html )
|
please test for slc7_ppc64le_gcc820 |
please test for slc7_aarch64_gcc820 |
The tests are being triggered in jenkins.
|
@smuzaffar is this
correct ? |
enable gpu |
@fwyzard , one needs to explicitly enable additional tests |
I am sorry but I do not understand what that means, concretely. If the latter, does it mean that @cmsbot will still run some basic tests ? Bottom line is, please test this PR for
Thank you! |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
please test |
The tests are being triggered in jenkins. |
+1 |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
+1 backport looks consistent with the update in master |
+externals |
This pull request is fully signed and it will be integrated in one of the next IB/CMSSW_11_0_X/master IBs (tests are also fine). This pull request will be automatically merged. |
Backport of #5418.
CUDA 10.1 Update 2 and later show a problem with running under MPS (NVIDIA Multi Process Server) or using CUDA Dynamic Parallelism within CMSSW.
While awaiting feedback from NVIDIA, the only solution seems to revert to the latest working version, which was 10.1 Update 1.
Drop the Nsight Compute and Nsigh System tools from the CUDA package, because they are released much more often as external packages.
Add support for CUDA on IBM Power architecture (ppc64le) on Linux.