Skip to content

SHMEM/SCOLL: Fix inplace reductions #8041

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

Merged
merged 1 commit into from
Sep 23, 2020

Conversation

brminich
Copy link
Member

According to shmem spec source and dest arrays can be the same for reductions:
The source and dest arrays may be the same array, but they may not be overlapping arrays.
If this is the case and MPI scoll is used, need to specify MPI_IN_PLACE instead of real source buffer for allreduce

Fixes openucx/ucx#5627

Signed-off-by: Mikhail Brinskii [email protected]

@ibm-ompi
Copy link

The IBM CI (PGI) build failed! Please review the log, linked below.

Gist: https://gist.github.com/ibm-ompi/88b480fb98dfd2bdac939f023e1f8fe6

@ibm-ompi
Copy link

The IBM CI (GNU/Scale) build failed! Please review the log, linked below.

Gist: https://gist.github.com/ibm-ompi/88b480fb98dfd2bdac939f023e1f8fe6

@ibm-ompi
Copy link

The IBM CI (XL) build failed! Please review the log, linked below.

Gist: https://gist.github.com/ibm-ompi/88b480fb98dfd2bdac939f023e1f8fe6

@jsquyres
Copy link
Member

bot:ibm:retest

@brminich
Copy link
Member Author

@yosefe, ok to merge?

Comment on lines 233 to 234
if (source == target) {
sbuf = MPI_IN_PLACE;
rbuf = source;
} else {
sbuf = (void *) source;
rbuf = target;
}

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

sbuf = (source == target) ? MPI_IN_PLACE : source
rbuf = target

Signed-off-by: Mikhail Brinskii <[email protected]>
@brminich brminich force-pushed the topic/shmem_scoll_fix branch from 5a89fd4 to dfe20e0 Compare September 23, 2020 07:07
@yosefe yosefe merged commit b532564 into open-mpi:master Sep 23, 2020
awlauria added a commit to awlauria/ompi that referenced this pull request Aug 13, 2021
This was already fixed on the master/v5.0.x branch by open-mpi#8041:
dfe20e0

This reverts commit 1ddcfcc.

Signed-off-by: Austen Lauria <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[mtt] oshmem_test, job has been aborted.
4 participants