Use phase barriers to work around the CUDA driver bug (#246) #247
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
* Utility methods to construct futures from buffers or cdata * Use NVML to check if the CUDA driver has a fix for GH nv-legate#374 * Use barriers at the beginning of NCCL tasks to work around a driver bug * Install Legion C util as it's now used by the core * Rename the parameter to match the definition * Use a CUDA driver call for the version check * Add some note on what the barrier is intended to do
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
Cherry-picked from
branch-22.07.Utility methods to construct futures from buffers or cdata
Use NVML to check if the CUDA driver has a fix for GH Check if ninja is available, fall back to gmake if not #374
Use barriers at the beginning of NCCL tasks to work around a driver bug
Install Legion C util as it's now used by the core
Rename the parameter to match the definition
Use a CUDA driver call for the version check
Add some note on what the barrier is intended to do