[SYCL] Lower queue::wait() to piQueueFinish when possible #4044
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 patch changes the logic of queue::wait() from waiting on each individual event in order of submission of their tasks to checking if each event's task has been enqueued, waiting for those that haven't been and calling piQueueFinish to take care of the rest. Notable exceptions to this new behaviour are host queues, queues that emulate out-of-order execution by creating multiple queues underneath and host task events, which are run on host regardless of the queue they are bound to.
sergey-semenov
requested review from
a team and
rbegam
and removed request for
a team
|
TBD: add unit tests. |
Any blocked command was being added to MPreparedHostDepsEvents which led to waiting for the dependency event before enqueueing the dependant command rather than passing the dependency to PI.
s-kanaev
previously approved these changes
Jul 6, 2021
s-kanaev
previously approved these changes
Jul 6, 2021
|
@sergey-semenov, please, fix pre-commit failures. |
Level Zero plugin has recently started to call wait on events during piEventRelease. With the queue::wait change we started to release some events during submit, prior to their completion. This introduces implicit wait on every submit that would produce such an event, which also interferes with level zero batching. Work around this by storing such events for level zero only to release them later.
@bader @vladimirlaz Done, the remaining failures are expected and covered by intel/llvm-test-suite#338 |
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Outdated suggestions cannot be applied.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.
This patch changes the logic of queue::wait() from waiting on each
individual event in order of submission of their tasks to checking if
each event's task has been enqueued, waiting for those that haven't been
and calling piQueueFinish to take care of the rest.
Notable exceptions to this new behaviour are host queues, queues that
emulate out-of-order execution by creating multiple queues underneath
and host task events, which are run on host regardless of the queue they
are bound to.