[svm][SYCL] In-order queue and native commands #101
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
Add the queue properties in_order and, if available, discard_events. The addition is steered by a CMake build option IN_ORDER_QUEUE. Set the default value to ON for NVIDIA and AMD backends and keep as OFF for other backends.
Use the wrapper functions from <top-dir>/infrastructure/SYCL.h to call either host_task or native command extensions when available. This improves performance with CUDA and HIP backends. Delete local copies of the SYCL.(h|cpp) files which were not used in the benchmark.
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Outdated suggestions cannot be applied.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.
Improve SYCL performance on CUDA and HIP backends with the two changes below. There is no functional change for Intel backends.
1. Add CMake option to use in-order queue
Add the queue properties
in_orderand, if available,discard_events. The addition is steered by a CMake build optionIN_ORDER_QUEUE. Set the default value to ON for NVIDIA and AMD backends and keep as OFF for other backends. The in-order queue corresponds more directly to how the CUDA/HIP variants of the benchmark are written since they use the default stream.2. Use extensions to submit native commands when available
Use the wrapper functions from
<top-dir>/infrastructure/SYCL.h(added in #95) to call eitherhost_taskor native command extensions when available. This improves performance with CUDA and HIP backends.Delete local copies of the SYCL.(h|cpp) files which were not used in the benchmark and clashed with the one used in this PR.