MPI_T Events by cchambreau · Pull Request #8057 · open-mpi/ompi

cchambreau · 2020-09-22T21:21:28Z

This PR includes Nathan Hjelm's reworking of PERUSE functionality to implement the MPI_T Events API as specified in the MPI 4.0 proposal. Subsequent minor changes to the MPI_T Events API were made and included here. The API and an initial set of events were tested against Marc-André Hermann's MEL tracing tool and other test cases. History, a description of the API and experience with tools can be found in "Enabling callback-driven runtime introspection via MPI_T."

ompiteam-bot · 2020-09-22T21:21:31Z

Can one of the admins verify this patch?

jsquyres · 2020-09-22T21:22:21Z

ok to test

gpaulsen · 2021-03-26T15:40:03Z

@cchambreau, Hi. Is this desirable for Open MPI v5.0 (not MPI-4.0 compliant)? If so please, rebase it and get it into master and then cherry-pick to v5.0.x branch after it's in master.

jsquyres

Can you rebase and fix all current conflicts?

Do we have any tests for this MPI_T event functionality?

ompi/mpi/tool/event_get_info.c

hppritcha · 2021-09-21T15:50:16Z

@cchambreau could you please rebase? thanks.

hppritcha · 2021-11-16T16:31:44Z

Howard will rebase and merge unless objections are raised.

ompi/mpi/tool/event_get_info.c

ompi/mpi/tool/event_handle_alloc.c

ompi/mpi/tool/event_handle_free.c

opal/mca/base/mca_base_event.c

ibm-ompi · 2021-12-23T23:55:09Z

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

Gist: https://gist.github.com/d182f369f64bb66a239cbf4b56b1e3ce

ibm-ompi · 2021-12-23T23:57:08Z

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

Gist: https://gist.github.com/20f1009df186bd52f322544e39a3595c

jsquyres · 2022-01-01T15:35:49Z

@cchambreau can you fix the conflicts on this PR? Thanks!

jsquyres

In addition to the other comments on this PR, please update this PR to use opal_clock_gettime() (and, if necessary, opal_clock_getres()) from #9798.

hppritcha · 2022-01-10T17:26:53Z

bot:ompi:retest

ompi/mpi/tool/event_callback_get_info.c

ompi/mpi/tool/event_get_info.c

ompi/mpi/tool/event_handle_alloc.c

ompi/mpi/tool/source_get_num.c

opal/mca/base/mca_base_event.h

opal/mca/base/mca_base_source.c

jsquyres · 2022-04-08T04:09:33Z

Hey @hppritcha Are there any plans to continue this work (and delete PERUSE, per #189)?

Signed-off-by: Howard Pritchard <howardp@lanl.gov>

new profile interface generation method Signed-off-by: Howard Pritchard <howardp@lanl.gov>

Signed-off-by: Howard Pritchard <howardp@lanl.gov>

by fixing some code Signed-off-by: Howard Pritchard <howardp@lanl.gov>

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

values. Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

Signed-off-by: Howard Pritchard <howardp@lanl.gov>

hppritcha · 2024-01-02T19:29:05Z

@cchambreau i'm trying to get this PR fixed up for merging so did a rebase and pushed to your repo.

hppritcha · 2024-04-24T18:22:06Z

Discussed some at the 2024 F2F. Not sure who really needs this. Joseph will ping scorep people about whether they may want/need this feature.

jsquyres · 2024-04-24T19:06:22Z

Is it really a question of who needs this?

I.e., isn't this part of conformance to the MPI standard?

hppritcha · 2024-04-29T18:02:22Z

Some tools folks were pinged about use of MPI_T events as part of Open MPI devel f2f meeting 4/2024. ScoreP developer said its on their todo list to use this feature in some future release. Tau may already use events if the MPI being used supports it. Possibly some unreleased version of MPI Advisor can use them.

kingshuk00 · 2024-05-23T15:55:23Z

It would be nice to have this standard-compliant interface from tools perspective.
I have posted another PR PR-6 on the source branch of this PR. This fixes a potential problem that is captured in this test events_callbacks.c. This should be there for correctness.

Turns out I forgot to run the pcvs checker when adding MPI_T event stubs in PR open-mpi#13086 and missed a couple of events related functions. Also it looks like these were not included in PR open-mpi#8057. With this patch, the PCVS MPI API checker passes for MPI 4.0 standard. The PCVS MPI API checker is described here https://dl.acm.org/doi/abs/10.1145/3615318.3615329 Signed-off-by: Howard Pritchard <howardp@lanl.gov>

hppritcha · 2026-01-08T21:54:31Z

this effort has been moved to #13133

jsquyres added enhancement MPI-4.0 Target: main labels Sep 22, 2020

gpaulsen added the Target: v5.0.x label Sep 29, 2020

gpaulsen added Target: v6.0.x and removed Target: v5.0.x labels Jan 28, 2021

jsquyres reviewed Jul 29, 2021

View reviewed changes

ompi/mpi/tool/event_get_info.c Outdated Show resolved Hide resolved

hppritcha self-assigned this Nov 16, 2021

hppritcha force-pushed the topic/mpi_t_events branch from 4b3c70d to 2400b8a Compare November 24, 2021 21:48

hppritcha added Target: v5.0.x and removed Target: v6.0.x labels Nov 30, 2021

jsquyres reviewed Dec 5, 2021

View reviewed changes

cchambreau mentioned this pull request Dec 24, 2021

New MPI_T Events tests and example open-mpi/ompi-tests-public#13

Merged

jsquyres mentioned this pull request Jan 1, 2022

RFC: opal_clock_gettime() and opal_clock_getres() #9798

Merged

jsquyres requested changes Jan 4, 2022

View reviewed changes

hppritcha force-pushed the topic/mpi_t_events branch 2 times, most recently from 3ed1cd9 to 14ccc8d Compare January 7, 2022 22:11

hppritcha force-pushed the topic/mpi_t_events branch from 14ccc8d to 9e35c26 Compare January 18, 2022 16:38

hppritcha requested a review from jsquyres January 19, 2022 02:23

jsquyres reviewed Jan 23, 2022

View reviewed changes

hppritcha and others added 16 commits December 21, 2023 15:32

mpi_t events: convert to new profiling approach

2152799

Signed-off-by: Howard Pritchard <howardp@lanl.gov>

mpi_t events fix some missing files for

7d81266

new profile interface generation method Signed-off-by: Howard Pritchard <howardp@lanl.gov>

pr review feedback

50b84fc

Signed-off-by: Howard Pritchard <howardp@lanl.gov>

pr feedback - switch to opal_clock_gettime

fb8a5f5

Signed-off-by: Howard Pritchard <howardp@lanl.gov>

suppress a compiler warning

f4ffec2

by fixing some code Signed-off-by: Howard Pritchard <howardp@lanl.gov>

pr feedback - remove assert

6cf3312

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

pr feedback - corrected comment references to events

12dca5d

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

pr feedback - added locking for setting dropped event handler

e6a0389

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

pr feedback - copyright consistency for all event files

7e617d4

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

Protect source data access with thread locking.

d91a7ee

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

Add OPAL_LIKELY

35e202a

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

Correct OSC event descriptions.

88c7717

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

Ensure consistency between ompi and opal callback safety requirement

9ac3f5f

values. Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

Ensure thread safety for incremented variables.

8853c60

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

Add function to lookup ompi datatype from opal datatype id.

750af0c

Signed-off-by: Chris Chambreau <chambreau1@llnl.gov>

fix compile problem after rebase

66b29d5

Signed-off-by: Howard Pritchard <howardp@lanl.gov>

hppritcha force-pushed the topic/mpi_t_events branch from b814b33 to 66b29d5 Compare January 2, 2024 19:28

hppritcha mentioned this pull request Feb 7, 2025

MPI_T: provide a stub implementation of events #13086

Merged

mahermanns mentioned this pull request Feb 17, 2025

Side Document defining MPI_T entities mpi-forum/mpi-issues#817

Open

kingshuk00 mentioned this pull request Mar 10, 2025

MPI_T Events #13133

Open

hppritcha mentioned this pull request Apr 15, 2025

MPI_T_Events: but wait there is more! #13197

Merged

hppritcha closed this Jan 8, 2026

Conversation

cchambreau commented Sep 22, 2020

Uh oh!

ompiteam-bot commented Sep 22, 2020

Uh oh!

jsquyres commented Sep 22, 2020

Uh oh!

gpaulsen commented Mar 26, 2021

Uh oh!

jsquyres left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

hppritcha commented Sep 21, 2021

Uh oh!

hppritcha commented Nov 16, 2021

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ibm-ompi commented Dec 23, 2021

Uh oh!

ibm-ompi commented Dec 23, 2021

Uh oh!

jsquyres commented Jan 1, 2022

Uh oh!

jsquyres left a comment

Choose a reason for hiding this comment

Uh oh!

hppritcha commented Jan 10, 2022

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

jsquyres commented Apr 8, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

hppritcha commented Jan 2, 2024

Uh oh!

hppritcha commented Apr 24, 2024

Uh oh!

jsquyres commented Apr 24, 2024

Uh oh!

hppritcha commented Apr 29, 2024

Uh oh!

kingshuk00 commented May 23, 2024

Uh oh!

hppritcha commented Jan 8, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

11 participants

jsquyres commented Apr 8, 2022 •

edited

Loading