[detailed][benchmark] demonstration of cuda memory footprint with multi-stream use case #3480
Conversation
This file contains hidden or 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
|
@TroyGarden has exported this pull request. If you are a Meta employee, you can view the originating Diff in D85399705. |
meta-cla
bot
added
the
CLA Signed
TroyGarden
changed the title
TroyGarden
changed the title
TroyGarden
added a commit
to TroyGarden/torchrec
that referenced
this pull request
Oct 24, 2025
…-pytorch#3480) Summary: Google Document: https://docs.google.com/document/d/1Odt6oJJgvPDeVSQmQqrl3iUAl2yu_HPUVLJ92EqDiTI Workplace Post: https://fb.workplace.com/groups/429376538334034/permalink/1488841649054179/ Differential Revision: D85399705
TroyGarden
force-pushed
the
export-D85399705
branch
from
dc2defd to
5c07535
Compare
TroyGarden
added a commit
to TroyGarden/torchrec
that referenced
this pull request
Oct 24, 2025
…-pytorch#3480) Summary: Google Document: https://docs.google.com/document/d/1Odt6oJJgvPDeVSQmQqrl3iUAl2yu_HPUVLJ92EqDiTI Workplace Post: https://fb.workplace.com/groups/429376538334034/permalink/1488841649054179/ Differential Revision: D85399705
TroyGarden
force-pushed
the
export-D85399705
branch
2 times, most recently
from
ff60cfd to
0ae7f55
Compare
TroyGarden
added a commit
to TroyGarden/torchrec
that referenced
this pull request
Oct 24, 2025
…-pytorch#3480) Summary: Google Document: https://docs.google.com/document/d/1Odt6oJJgvPDeVSQmQqrl3iUAl2yu_HPUVLJ92EqDiTI Workplace Post: https://fb.workplace.com/groups/429376538334034/permalink/1488841649054179/ # context * high-level design and technical discussions are in the document/post * this diff added three benchmark jobs to demonstrate the memory footprint in multi-stream vs single-stream scenarios * other changes in the benchmark function: **a**. make reset_accumulated_memory_stats default to True **b**. call `torch.cuda.empty_cache()` before the memory snapshot so that the snapshot won't include the residual effects from the previous benchmark runs * benchmark stats |name| GPU Runtime|CPU Runtime|GPU Peak Memory **alloc**|GPU Peak Memory **reserved**| CPU Peak RSS| |--|--| |single_stream_memory|156.62 ms|231.05 ms|5.06 GB|5.11 GB|1.68 GB| |multi_stream_memory|144.52 ms|175.61 ms|5.06 GB |**10.13 GB** | 1.71 GB| |multi_stream_optimized|145.66 ms |232.76 ms | 5.06 GB |5.11 GB | 1.70 GB| Differential Revision: D85399705
…-pytorch#3480) Summary: Google Document: https://docs.google.com/document/d/1Odt6oJJgvPDeVSQmQqrl3iUAl2yu_HPUVLJ92EqDiTI Workplace Post: https://fb.workplace.com/groups/429376538334034/permalink/1488841649054179/ # context * high-level design and technical discussions are in the document/post * this diff added three benchmark jobs to demonstrate the memory footprint in multi-stream vs single-stream scenarios * other changes in the benchmark function: **a**. make reset_accumulated_memory_stats default to True **b**. call `torch.cuda.empty_cache()` before the memory snapshot so that the snapshot won't include the residual effects from the previous benchmark runs * benchmark stats |name| GPU Runtime|CPU Runtime|GPU Peak Memory **alloc**|GPU Peak Memory **reserved**| CPU Peak RSS| |--|--| |single_stream_memory|**158.58 ms**|233.79 ms|5.06 GB|5.11 GB|1.70 GB| |multi_stream_memory|145.43 ms|241.68 ms|5.06 GB |**10.13 GB** | 1.70 GB| |multi_stream_optimized|146.98 ms |244.72 ms |5.06 GB |5.11 GB | 1.70 GB| Reviewed By: spmex Differential Revision: D85399705
TroyGarden
force-pushed
the
export-D85399705
branch
from
0ae7f55 to
e2dbe39
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Differential Revision: D85399705