Improve SYCL reduction performance: RangePolicy #6264
Merged
Conversation
This file contains 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
|
Only |
nliber
approved these changes
Jul 12, 2023
dalg24
reviewed
Jul 13, 2023
masterleinad
force-pushed
the
improve_reduction_performance_sycl_1
branch
from
9ac49e9
to
c9573a6
Compare
dalg24
approved these changes
Jul 17, 2023
|
SYCL CI is passing. |
|
Unrelated failure to launch one CUDA build |
Closed
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Outdated suggestions cannot be applied.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.
Part of #6035.
This limits the number of workgroups, effectively processing multiple work items per thread, and increases the maximum workgroup size for Intel GPUs. Technically,
sycl::info::kernel_device_specific::work_group_sizeshould be the maximum usable value for the workgroup size but it turns out I could still choosesycl::info::device::max_work_group_sizeand got better performance.Also, fix matching
use_shuffle_based_algorithm(is the reference not a pointer) withReducerType::static_value_size()(0 for array reductions).