[CUBLAS, CUDNN] Support dynamic batch size #7194
Merged
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
Contributor
|
Thanks @masahi @kevinthesun |
tkonolige
pushed a commit
to tkonolige/incubator-tvm
that referenced
this pull request
Jan 11, 2021
* support cudnn and cublas on dynamic batch size * added test for cublas * add comment on algo choice
masahi
added a commit
to masahi/tvm
that referenced
this pull request
Jan 14, 2021
* support cudnn and cublas on dynamic batch size * added test for cublas * add comment on algo choice
TusharKanekiDey
pushed a commit
to TusharKanekiDey/tvm
that referenced
this pull request
Jan 20, 2021
* support cudnn and cublas on dynamic batch size * added test for cublas * add comment on algo choice
trevor-m
pushed a commit
to neo-ai/tvm
that referenced
this pull request
Jan 21, 2021
* support cudnn and cublas on dynamic batch size * added test for cublas * add comment on algo choice
electriclilies
pushed a commit
to electriclilies/tvm
that referenced
this pull request
Feb 18, 2021
* support cudnn and cublas on dynamic batch size * added test for cublas * add comment on algo choice
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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Outdated suggestions cannot be applied.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.
Similar to what's done in #6598, updated cublas and cudnn offload to enable them to take on dynamic batch size inputs.
Especially cublas is required for reasonable performance on PyTorch Faster RCNN / MaskRCNN, due to a large, dynamic batch dense layer they have.
please review @zhiics @comaniac @junrushao1994