Remove dependency on libtensorflow_framework.so. #27029
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
Remove libtensorflow_framework.so as a build product. This should reduce toolchain size.
|
@swift-ci Please test tensorflow |
|
@swift-ci Please clean test tensorflow |
pschuh
approved these changes
Sep 4, 2019
|
Please hold off on merging until toolchain building is tested locally. |
|
Toolchain building succeeded on macOS. |
dan-zheng
added a commit
to dan-zheng/swift
that referenced
this pull request
Sep 5, 2019
Bazel produces multiple TensorFlow library artifacts: ``` $ ls -alh tensorflow/bazel-bin/tensorflow 18B libtensorflow.so -> libtensorflow.so.1 23B libtensorflow.so.1 -> libtensorflow.so.1.14.0 277M libtensorflow.so.1.14.0 ``` Previously, TensorFlow libraries were copied using `cp -a` via a glob pattern. For some reason, symlinks were not preserved: ``` $ ls -alh <TOOLCHAIN_BEFORE>.xctoolchain/usr/lib/swift/macosx 18B libtensorflow.so -> libtensorflow.so.1 277M libtensorflow.so.1 277M libtensorflow.so.1.14.0 # duplicate library ``` Now, rather than copying all libraries via glob, `copy_file_preserving_symlinks` copies exactly the necessary source files to the destination while preserving symlinks: ``` $ ls -alh <TOOLCHAIN_AFTER>.xctoolchain/usr/lib/swift/macosx 18B libtensorflow.so -> libtensorflow.so.1 277M libtensorflow.so.1 ``` This combined with removing libtensorflow_framework.so dependency (swiftlang#27029) led to a macOS toolchain size reduction from 4.91 GB to 3.45 GB.
pschuh
pushed a commit
that referenced
this pull request
Sep 6, 2019
…27032) Bazel produces multiple TensorFlow library artifacts: ``` $ ls -alh tensorflow/bazel-bin/tensorflow 18B libtensorflow.so -> libtensorflow.so.1 23B libtensorflow.so.1 -> libtensorflow.so.1.14.0 277M libtensorflow.so.1.14.0 ``` Previously, TensorFlow libraries were copied using `cp -a` via a glob pattern. For some reason, symlinks were not preserved: ``` $ ls -alh <TOOLCHAIN_BEFORE>.xctoolchain/usr/lib/swift/macosx 18B libtensorflow.so -> libtensorflow.so.1 277M libtensorflow.so.1 277M libtensorflow.so.1.14.0 # duplicate library ``` Now, rather than copying all libraries via glob, `copy_file_preserving_symlinks` copies exactly the necessary source files to the destination while preserving symlinks: ``` $ ls -alh <TOOLCHAIN_AFTER>.xctoolchain/usr/lib/swift/macosx 18B libtensorflow.so -> libtensorflow.so.1 277M libtensorflow.so.1 ``` This combined with removing libtensorflow_framework.so dependency (#27029) led to a macOS toolchain size reduction from 4.91 GB to 3.45 GB.
dan-zheng
added a commit
that referenced
this pull request
Sep 18, 2019
Remove libtensorflow_framework.so as a build product. This reduces toolchain size. Follow-up attempt to #20898.
dan-zheng
added a commit
to dan-zheng/swift
that referenced
this pull request
Sep 18, 2019
Bazel produces multiple TensorFlow library artifacts: ``` $ ls -alh tensorflow/bazel-bin/tensorflow 18B libtensorflow.so -> libtensorflow.so.1 23B libtensorflow.so.1 -> libtensorflow.so.1.14.0 277M libtensorflow.so.1.14.0 ``` Previously, TensorFlow libraries were copied via `cp -p`. This did not preserve symlinks, leading to library duplication: ``` $ ls -alh <TOOLCHAIN_BEFORE>.xctoolchain/usr/lib/swift/macosx 18B libtensorflow.so -> libtensorflow.so.1 277M libtensorflow.so.1 277M libtensorflow.so.1.14.0 # duplicate library ``` Now, `cp -a` is used, which preserves symlinks: ``` $ ls -alh <TOOLCHAIN_AFTER>.xctoolchain/usr/lib/swift/macosx 18B libtensorflow.so -> libtensorflow.so.1 277M libtensorflow.so.1 ``` This combined with removing libtensorflow_framework.so dependency (swiftlang#27029) led to a macOS toolchain size reduction from 3.99 GB to 3.62 GB on the tensorflow-0.5 branch.
dan-zheng
added a commit
that referenced
this pull request
Sep 18, 2019
) Bazel produces multiple TensorFlow library artifacts: ``` $ ls -alh tensorflow/bazel-bin/tensorflow 18B libtensorflow.so -> libtensorflow.so.1 23B libtensorflow.so.1 -> libtensorflow.so.1.14.0 277M libtensorflow.so.1.14.0 ``` Previously, TensorFlow libraries were copied via `cp -p`. This did not preserve symlinks, leading to library duplication: ``` $ ls -alh <TOOLCHAIN_BEFORE>.xctoolchain/usr/lib/swift/macosx 18B libtensorflow.so -> libtensorflow.so.1 277M libtensorflow.so.1 277M libtensorflow.so.1.14.0 # duplicate library ``` Now, `cp -a` is used, which preserves symlinks: ``` $ ls -alh <TOOLCHAIN_AFTER>.xctoolchain/usr/lib/swift/macosx 18B libtensorflow.so -> libtensorflow.so.1 277M libtensorflow.so.1 ``` This combined with removing libtensorflow_framework.so dependency (#27029) led to a macOS toolchain size reduction from 3.99 GB to 3.62 GB on the tensorflow-0.5 branch.
dan-zheng
added a commit
that referenced
this pull request
Sep 18, 2019
) Bazel produces multiple TensorFlow library artifacts: ``` $ ls -alh tensorflow/bazel-bin/tensorflow 18B libtensorflow.so -> libtensorflow.so.1 23B libtensorflow.so.1 -> libtensorflow.so.1.14.0 277M libtensorflow.so.1.14.0 ``` Previously, TensorFlow libraries were copied via `cp -p`. This did not preserve symlinks, leading to library duplication: ``` $ ls -alh <TOOLCHAIN_BEFORE>.xctoolchain/usr/lib/swift/macosx 18B libtensorflow.so -> libtensorflow.so.1 277M libtensorflow.so.1 277M libtensorflow.so.1.14.0 # duplicate library ``` Now, `cp -a` is used, which preserves symlinks: ``` $ ls -alh <TOOLCHAIN_AFTER>.xctoolchain/usr/lib/swift/macosx 18B libtensorflow.so -> libtensorflow.so.1 277M libtensorflow.so.1 ``` This combined with removing libtensorflow_framework.so dependency (#27029) led to a macOS toolchain size reduction from 3.99 GB to 3.62 GB on the tensorflow-0.5 branch.
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.
Remove
libtensorflow_framework.soas a build product.This should reduce toolchain size.
Follow-up attempt to #20898.