Skip to content
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

v2.x: disable hwloc CUDA support when relevant #4250

Merged

Conversation

jsquyres
Copy link
Member

Refs #4248.

@jsquyres jsquyres added the bug label Sep 21, 2017
@jsquyres jsquyres added this to the v2.1.3 milestone Sep 21, 2017
@jjhursey
Copy link
Member

bot:ibm:xl:retest

@open-mpi open-mpi deleted a comment from ibm-ompi Sep 21, 2017
@jsquyres
Copy link
Member Author

Working through issues identified on #4251.

@jsquyres jsquyres force-pushed the pr/v2.x/disable-hwloc-cuda-when-relevant branch 2 times, most recently from b59b67e to b68443f Compare September 25, 2017 15:48
@jsquyres
Copy link
Member Author

@ggouaillardet New commit added to this v2.x PR.

jsquyres and others added 3 commits October 10, 2017 09:59
(originally cherry picked from master, but since the hwloc component
was updated between master and v2.0.x but are the same between v2.0.x
and v2.x, I cherry picked from the v2.0.x branch for the v2.x branch)

Signed-off-by: Jeff Squyres <[email protected]>
(cherry picked from commit f5d51dc)
(cherry picked from commit 3b01bbe)
Open MPI doesn't use GL or OpenCL OS devices, so just disable them in
hwloc.

(originally cherry picked from master, but since the hwloc component
was updated between master and v2.0.x but are the same between v2.0.x
and v2.x, I cherry picked from the v2.0.x branch for the v2.x branch)

Signed-off-by: Jeff Squyres <[email protected]>
(cherry picked from commit 84a721d)
(cherry picked from commit ca0973a)
Be a little more deliberate about convering OMPI's --with-cuda CLI
value to hwloc's --enable-cuda configure option.

Also, unconditionally disable hwloc NVML support (because Open MPI is
not currently using it).

Signed-off-by: Jeff Squyres <[email protected]>
(cherry picked from commit 2ec2a32)
@jsquyres jsquyres force-pushed the pr/v2.x/disable-hwloc-cuda-when-relevant branch from 86114b8 to 7529579 Compare October 10, 2017 16:59
There is no usage of CUDA hwloc objects in the v2.x branch, and
linking in CUDA can cause problems (per
open-mpi#4257 (comment)).

Partially cherry-picked from c341b53.

Signed-off-by: Jeff Squyres <[email protected]>
@jsquyres jsquyres force-pushed the pr/v2.x/disable-hwloc-cuda-when-relevant branch from 7529579 to afe9abc Compare October 10, 2017 17:15
@jsquyres
Copy link
Member Author

@hppritcha Updated this v2.x PR. Can you review?

@jjhursey
Copy link
Member

bot:ibm:gnu:retest

@hppritcha
Copy link
Member

had to kill off armv8 testing.
bot:lanl:retest

@jsquyres jsquyres merged commit d32b0fa into open-mpi:v2.x Oct 12, 2017
@jsquyres jsquyres deleted the pr/v2.x/disable-hwloc-cuda-when-relevant branch October 12, 2017 14:40
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

6 participants