Roctracer: Device ID is off by 2, manually start from 0 #925
Conversation
This file contains 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
|
This pull request was exported from Phabricator. Differential Revision: D56951239 |
aaronenyeshi
added a commit
to aaronenyeshi/kineto
that referenced
this pull request
May 3, 2024
Summary: Although hipGetDeviceProperties shows 8 devices enumerating from 0 to 7, when using roctracer_record_t, the record->device_id enumerates from 2 to 9. Manually enumerate from 0-7 by subtracting 2, and opening a bug report in ROCm/roctracer. Differential Revision: D56951239
aaronenyeshi
force-pushed
the
export-D56951239
branch
from
c8ee6ab
to
6bf4cf7
Compare
|
This pull request was exported from Phabricator. Differential Revision: D56951239 |
aaronenyeshi
added a commit
to aaronenyeshi/kineto
that referenced
this pull request
May 3, 2024
Summary: Although hipGetDeviceProperties shows 8 devices enumerating from 0 to 7, when using roctracer_record_t, the record->device_id enumerates from 2 to 9. Manually enumerate from 0-7 by subtracting 2, and opened a bug report in ROCm/roctracer: ROCm/roctracer#98. Test Plan: CI and ran locally. If using hipGetDeviceProperties, the ids start from 0 to 7, but Roctracer's roctracer_record_t enumerates devices from 2-7. Tested various combinations. ## Ran on GPU 0-7: Before: {F1503734690} {F1503734811} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714765825/localhost/libkineto_activities_1679469.json.gz&bucket=gpu_traces After: {F1503735114} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714767517/localhost/libkineto_activities_3723929.json.gz&bucket=gpu_traces ## Ran on GPU 0: Before, shows GPU 2: {F1503735699} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714766546/localhost/libkineto_activities_2685136.json.gz&bucket=gpu_traces After, shows GPU 0: {F1503735909} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714768432/localhost/libkineto_activities_764017.json.gz&bucket=gpu_traces ## Ran on GPU 7: Before, shows GPU 9: {F1503736130} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714767155/localhost/libkineto_activities_3159876.json.gz&bucket=gpu_traces After, shows GPU 7: {F1503736726} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714768581/localhost/libkineto_activities_1006961.json.gz&bucket=gpu_traces Differential Revision: D56951239 Pulled By: aaronenyeshi
aaronenyeshi
force-pushed
the
export-D56951239
branch
from
6bf4cf7
to
c7a300c
Compare
|
This pull request was exported from Phabricator. Differential Revision: D56951239 |
aaronenyeshi
changed the title
Summary: Although hipGetDeviceProperties shows 8 devices enumerating from 0 to 7, when using roctracer_record_t, the record->device_id enumerates from 2 to 9. Manually enumerate from 0-7 by subtracting 2, and opened a bug report in ROCm/roctracer: ROCm/roctracer#98. Test Plan: CI and ran locally. If using hipGetDeviceProperties, the ids start from 0 to 7, but Roctracer's roctracer_record_t enumerates devices from 2-7. Tested various combinations. ## Ran on GPU 0-7: Before: {F1503734690} {F1503734811} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714765825/localhost/libkineto_activities_1679469.json.gz&bucket=gpu_traces After: {F1503735114} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714767517/localhost/libkineto_activities_3723929.json.gz&bucket=gpu_traces ## Ran on GPU 0: Before, shows GPU 2: {F1503735699} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714766546/localhost/libkineto_activities_2685136.json.gz&bucket=gpu_traces After, shows GPU 0: {F1503735909} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714768432/localhost/libkineto_activities_764017.json.gz&bucket=gpu_traces ## Ran on GPU 7: Before, shows GPU 9: {F1503736130} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714767155/localhost/libkineto_activities_3159876.json.gz&bucket=gpu_traces After, shows GPU 7: {F1503736726} https://www.internalfb.com/intern/perfdoctor/trace_view?filepath=tree/traces/dynocli/0/1714768581/localhost/libkineto_activities_1006961.json.gz&bucket=gpu_traces Differential Revision: D56951239 Pulled By: aaronenyeshi
aaronenyeshi
force-pushed
the
export-D56951239
branch
from
c7a300c
to
85cece9
Compare
|
This pull request was exported from Phabricator. Differential Revision: D56951239 |
houseroad
suggested changes
May 6, 2024
|
Closing this as we may have a better fix via: #926. prototyping it. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Summary:
Although hipGetDeviceProperties shows 8 devices enumerating from 0 to 7, when using roctracer_record_t, the record->device_id enumerates from 2 to 9.
Manually enumerate from 0-7 by subtracting 2, and opened a bug report in ROCm/roctracer: ROCm/roctracer#98.
Differential Revision: D56951239