Avoid h2d - d2h roundtrip when using unflatten
#861
Merged
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Outdated suggestions cannot be applied.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.
Description
unflatten
converts itslengths
argument to a NumPy array, because CuPy'ssplit
function requires lengths to be in CPU memory. However, in various places in Thinc, we copy the lengths array to GPU memory when CupyOps is used. This results in an unnecessary roundtrip of the lengths array (host to device -> device to host). One of these roundtrips (arraylist2array
) showed up in profiles of the biaffine parser.This change fixes some length array allocations to avoid the round trip.
Types of change
Performance improvement
Checklist