Reduce unnecesary function call and use zip to transpose 2d list #2796
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
facebook-github-bot
added
the
CLA Signed
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D70960056 |
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D70960056 |
elokrainz
force-pushed
the
export-D70960056
branch
from
4431005 to
ebc17d8
Compare
…eta-pytorch#2796) Summary: This diff aims to improve the performance of the `_set_sharding_context_post_a2a` in `embedding_sharding` module in TorchRec by reducing unnecessary function calls. _set_sharding_context_post_a2a_previous-KJT_len:100-keys:10 | Runtime (P90): 0.45 ms _set_sharding_context_post_a2a-KJT_len:100-keys:10 | Runtime (P90): 0.28 ms _set_sharding_context_post_a2a_previous-KJT_len:100-keys:100 | Runtime (P90): 2.47 ms _set_sharding_context_post_a2a-KJT_len:100-keys:100 | Runtime (P90): 0.89 ms _set_sharding_context_post_a2a_previous-KJT_len:1000-keys:10 | Runtime (P90): 4.55 ms _set_sharding_context_post_a2a-KJT_len:1000-keys:10 | Runtime (P90): 2.73 ms _set_sharding_context_post_a2a_previous-KJT_len:1000-keys:100 | Runtime (P90): 24.49 ms _set_sharding_context_post_a2a-KJT_len:1000-keys:100 | Runtime (P90): 8.68 ms _set_sharding_context_post_a2a_previous-KJT_len:10000-keys:10 | Runtime (P90): 46.85 ms _set_sharding_context_post_a2a-KJT_len:10000-keys:10 | Runtime (P90): 28.00 ms _set_sharding_context_post_a2a_previous-KJT_len:10000-keys:100 | Runtime (P90): 243.57 ms _set_sharding_context_post_a2a-KJT_len:10000-keys:100 | Runtime (P90): 89.14 ms Previous trace: about 4-5ms {F1975959472} Now: less than 4ms, some even less than 0.2ms, with much less operations {F1975959630} {F1975959609} Test on IG TAB ESR TTSN model: NE trending is same: {F1975957602} {F1975957605} {F1975957635} Differential Revision: D70960056 Privacy Context Container: L1292699
elokrainz
force-pushed
the
export-D70960056
branch
from
ebc17d8 to
03c30e4
Compare
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D70960056 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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:
This diff aims to improve the performance of the
_set_sharding_context_post_a2ainembedding_shardingmodule in TorchRec by reducing unnecessary function calls and usingzip_longestto transpose 2D jagged lists.