[AutoDiff] Speed up compile time by 3.5x by caching derivative function types. #29590
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
|
@swift-ci please test tensorflow |
|
@swift-ci please test tensorflow linux gpu |
dan-zheng
approved these changes
Feb 1, 2020
|
Linux GPU CI failures are temporarily expected. Merging now so new toolchains can be built with the compilation time fix. |
|
I believe that this change may have regressed the Windows builds: |
rxwei
added a commit
to rxwei/swift
that referenced
this pull request
Feb 1, 2020
|
My apologies. I just sent PR #29592 your way. |
compnerd
added a commit
that referenced
this pull request
Feb 2, 2020
[AutoDiff] Fix windows build failure introduced in #29590.
|
Very nice drop on our graph of completime!!
|
|
Nice graph. This shows exactly when the slowdown occurred: It's when we switched to |
dan-zheng
added a commit
to dan-zheng/swift
that referenced
this pull request
Feb 20, 2020
dan-zheng
added a commit
that referenced
this pull request
Feb 20, 2020
Upstream #29590: cache `SILFunctionType::getAutoDiffDerivativeFunctionType` results.
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Outdated suggestions cannot be applied.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.
Compiling packages that use AutoDiff is currently extremely slow. For example, compiling tensorflow/swift-apis takes ~5 minutes on a slower machine. This is largely due to calls to
SILFunctionType::getAutoDiffDerivativeFunctionTypein SIL verifier, which spends a long time computing generic signatures with GSB.This PR improves this situation by caching computed derivative function types in
ASTContext. This resulted in a 1.6x speedup in running the AutoDiff test suite, and a 3.5x speedup in compiling tensorflow/swift-apis.Future investigations:
getAutoDiffDerivativeFunctionGenericSignature.SILFunctionType::getAutoDiffTransposeFunctionType().Here are the results building tensorflow/swift-apis with
-j56.Before:
After:
Resolves TF-1133.