[NCCL] Tests for WorkNCCL::wait with Timeouts #40947
Closed
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
This PR adds tests for work-level timeouts in WorkNCCL objects. We kick off an allgather operation that waits for 1000ms before actually starting computation. We wait on completion of this allgather op with a timeout of 250ms, expecting the operation to timeout and throw a runtime error. Differential Revision: [D22173101](https://our.internmc.facebook.com/intern/diff/D22173101/) [ghstack-poisoned]
This was referenced Jul 2, 2020
💊 CI failures summary and remediationsAs of commit 3e25ba8 (more details on the Dr. CI page):
🕵️ 3 new failures recognized by patternsThe following CI failures do not appear to be due to upstream breakages:
|
This PR adds tests for work-level timeouts in WorkNCCL objects. We kick off an allgather operation that waits for 1000ms before actually starting computation. We wait on completion of this allgather op with a timeout of 250ms, expecting the operation to timeout and throw a runtime error. Differential Revision: [D22173101](https://our.internmc.facebook.com/intern/diff/D22173101/) [ghstack-poisoned]
This PR adds tests for work-level timeouts in WorkNCCL objects. We kick off an allgather operation that waits for 1000ms before actually starting computation. We wait on completion of this allgather op with a timeout of 250ms, expecting the operation to timeout and throw a runtime error. Differential Revision: [D22173101](https://our.internmc.facebook.com/intern/diff/D22173101/) [ghstack-poisoned]
mrshenli
approved these changes
Jul 7, 2020
| // First we reset the NCCL_BLOCKING_WAIT environment variable to prevent | ||
| // any unwanted side-effects. We must do this at all exit-points for this | ||
| // function. | ||
| setenv(c10d::NCCL_BLOCKING_WAIT, originalBlockingWait, 1); |
Contributor
There was a problem hiding this comment.
nit (please feel free to ignore): does it worth it to make this a guard?
Contributor
There was a problem hiding this comment.
Good point. It's kind of overhead to restore original env value
| setenv(c10d::NCCL_BLOCKING_WAIT, originalBlockingWait, 1); | ||
| // If no exception, that is also an error since we expect the test.wait call | ||
| // to timeout and throw. | ||
| throw std::runtime_error("BOOM!"); |
Contributor
There was a problem hiding this comment.
Can we make the message more informative here? Same for the above two throws.
Member
Author
There was a problem hiding this comment.
I was actually planning on cleaning up these messages, but just followed the same error messages as the already existing tests for now (they all do throw std::runtime_error("BOOM!") for some reason...). I'm planning on putting up another PR that fixes all of these.
This PR adds tests for work-level timeouts in WorkNCCL objects. We kick off an allgather operation that waits for 1000ms before actually starting computation. We wait on completion of this allgather op with a timeout of 250ms, expecting the operation to timeout and throw a runtime error. Differential Revision: [D22173101](https://our.internmc.facebook.com/intern/diff/D22173101/) [ghstack-poisoned]
This PR adds tests for work-level timeouts in WorkNCCL objects. We kick off an allgather operation that waits for 1000ms before actually starting computation. We wait on completion of this allgather op with a timeout of 250ms, expecting the operation to timeout and throw a runtime error. Differential Revision: [D22173101](https://our.internmc.facebook.com/intern/diff/D22173101/) [ghstack-poisoned]
jiayisuse
approved these changes
Jul 15, 2020
| // First we reset the NCCL_BLOCKING_WAIT environment variable to prevent | ||
| // any unwanted side-effects. We must do this at all exit-points for this | ||
| // function. | ||
| setenv(c10d::NCCL_BLOCKING_WAIT, originalBlockingWait, 1); |
Contributor
There was a problem hiding this comment.
Good point. It's kind of overhead to restore original env value
This PR adds tests for work-level timeouts in WorkNCCL objects. We kick off an allgather operation that waits for 1000ms before actually starting computation. We wait on completion of this allgather op with a timeout of 250ms, expecting the operation to timeout and throw a runtime error. Differential Revision: [D22173101](https://our.internmc.facebook.com/intern/diff/D22173101/) [ghstack-poisoned]
This PR adds tests for work-level timeouts in WorkNCCL objects. We kick off an allgather operation that waits for 1000ms before actually starting computation. We wait on completion of this allgather op with a timeout of 250ms, expecting the operation to timeout and throw a runtime error. Differential Revision: [D22173101](https://our.internmc.facebook.com/intern/diff/D22173101/) [ghstack-poisoned]
Contributor
|
This pull request has been merged in 01dcef2. |
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.
Stack from ghstack:
This PR adds tests for work-level timeouts in WorkNCCL objects. We kick off an allgather operation that waits for 1000ms before actually starting computation. We wait on completion of this allgather op with a timeout of 250ms, expecting the operation to timeout and throw a runtime error.
Differential Revision: D22173101