Feat/op fusion decorator #939
Merged
Conversation
This file contains 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
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## main #939 +/- ##
==========================================
+ Coverage 86.44% 87.25% +0.81%
==========================================
Files 470 486 +16
Lines 44326 49666 +5340
==========================================
+ Hits 38317 43336 +5019
- Misses 6009 6330 +321
☔ View full report in Codecov by Sentry. |
louisfd
approved these changes
Nov 10, 2023
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Outdated suggestions cannot be applied.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.
Burn Fusion
A new backend decorator to perform graph optimizations on a stream of operations. Backends that want to support this functionality can implement the new
FusionBackendtrait in addition to theBackendtrait.This allows the backend to provide a list of potential optimizations to be performed instead of executing a list of normal backend operations.
For now, only the Wgpu backend implements the new trait, but it allows to actually test that the new backend works even without optimizations.
Following PRs will bring new optimizations to the Wgpu backend 🔥
Overhead ?
There is some framework overhead introduced by this decorator, but since most backends are async, it shouldn't impact the speed of execution. In fact, I enabled the new backend decorator on the text classification example and the training time isn't longer than before.
The actual overhead introduced is very minimal and is constant per operation. We will introduce a cache to reduce the overhead of fusion operation implementations #935