Optimized ONNX Transform via Class Merging and Thread Pooling #546
+90
−74
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
abhishek-singh591
requested review from
quic-rishinr,
ochougul,
quic-hemagnih and
quic-amitraj
as code owners
ochougul
requested changes
Sep 1, 2025
abhishek-singh591
force-pushed
the
optimized_onnx_tranform
branch
from
2df6780 to
a528b29
Compare
ochougul
requested changes
Sep 4, 2025
abhishek-singh591
force-pushed
the
optimized_onnx_tranform
branch
2 times, most recently
from
00fcaf2 to
9b9c41d
Compare
Signed-off-by: abhishek-singh591 <[email protected]>
…added removed comments Signed-off-by: abhishek-singh591 <[email protected]>
…added removed comments Signed-off-by: abhishek-singh591 <[email protected]>
Signed-off-by: abhishek-singh591 <[email protected]>
Signed-off-by: abhishek-singh591 <[email protected]>
Signed-off-by: abhishek-singh591 <[email protected]>
abhishek-singh591
force-pushed
the
optimized_onnx_tranform
branch
from
e83ac9e to
f64b429
Compare
quic-rishinr
requested changes
Sep 10, 2025
| if not apply_clip and not apply_split: | ||
| warnings.warn("Both apply_clip and apply_split are False. Skipping transformation.") | ||
| return model, False | ||
|
|
||
| external_data_helper.load_external_data_for_model(model, onnx_base_dir) |
There was a problem hiding this comment.
Even though combining the transform save time in this case, it also reduces the flexibility we have over multiple transforms. In future if we need to add more transforms the condition would become more complex and if its a new transform would need to load the tensors again. I have added few changes as part #538 of the memory clean and reducing the peak memory usage. Can you check if the same concepts can be used here?
There was a problem hiding this comment.
Yeah, its kind off tradeoff between time and flexibility, let me check that will get back to you.
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.
Optimized ONNX Transform via Class Merging and Thread Pooling
This PR follows up on #539 – Optimized ONNX transform class via multithreading.
It merges the FP16 and Split ONNX transform classes into a single implementation to eliminate redundant tensor loading and iteration. Additionally, the transform logic has been refactored to use a thread pool, replacing the previous sequential loop to parallelize tensor operations.
Performance Benchmarks:-