Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Overhaul TravisCI #4207

Merged
merged 1 commit into from
May 25, 2016
Merged

Overhaul TravisCI #4207

merged 1 commit into from
May 25, 2016

Conversation

lukeyeager
Copy link
Contributor

@lukeyeager lukeyeager commented May 24, 2016

TODO:

  • Don't install the full CUDA toolkit

wontfix:

@lukeyeager
Copy link
Contributor Author

I was going to try to fix the Python3 + raw Makefile build, but I realized that we weren't testing it before. So I'll leave that for someone else to fix. But I'm skipping it more explicitly now.

LINE "WITH_PYTHON_LAYER := 1"

if $WITH_PYTHON3 ; then
# XXX doesn't work
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For notes like this we like to use TODO(handle) like the TODO(shelhamer) you see in the Caffe code. XXX isn't currently used in the project for highlighting comments.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok. Want me to assign it to anyone in particular?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh, I read this as the author of the TODO, and not the assignee, so TODO(lukeyeager) is fine. Don't worry, it's not a promise to do it yourself haha. Although versioning records the line author it's nice to have a quick reference for who thought of the TODO in case there are later questions or you want to search for commits by the same author for context.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

done

@lukeyeager
Copy link
Contributor Author

I've removed the protobuf3 build cache:

  1. The cuDNN build is the long pole anyway
  2. Very few people would know what's going on if/when the cache gets screwed up, and it requires a @BVLC member to clear the cache. Not worth the trouble in my opinion.
  3. We would need to decide a reasonable timeout limit greater than the default of 5 minutes if caching was going to be useful (see Migrate to new Travis-CI container infrastructure #3798 (comment)).

* Run on Ubuntu 14.04
* Test cuDNN builds
* Build with OpenBLAS

NOTE: Python3 build only works with CMake
@lukeyeager
Copy link
Contributor Author

I've done all I plan to do for this PR. Please let me know if other changes are necessary.

@flx42
Copy link
Contributor

flx42 commented May 25, 2016

Nice, since you trimmed the CUDA dependencies, is it building faster now?

@lukeyeager
Copy link
Contributor Author

Nice, since you trimmed the CUDA dependencies, is it building faster now?

The build didn't speed up, but installing dependencies doesn't take as long. As expected.

@shelhamer shelhamer merged commit 5a45b87 into BVLC:master May 25, 2016
@shelhamer
Copy link
Member

Thanks for the CI care @lukeyeager! These are welcome improvements.

@lukeyeager lukeyeager deleted the bvlc/travis-overhaul branch May 25, 2016 21:48
@lukeyeager
Copy link
Contributor Author

Sure thing. Can't wait to solve #3351 so we can cut the test matrix in half.

fxbit pushed a commit to Yodigram/caffe that referenced this pull request Sep 1, 2016
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Travis should check that USE_CUDNN builds
3 participants