-
Notifications
You must be signed in to change notification settings - Fork 752
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[SYCL] Extend scope of scheduler bypass to safe to bypass events #16735
base: sycl
Are you sure you want to change the base?
[SYCL] Extend scope of scheduler bypass to safe to bypass events #16735
Conversation
Alexandr-Konovalov
commented
Jan 22, 2025
- Aling scheduler bypass conditions for CG and for memory ops.
- Connect an event returned from scheduler bypass case with dependent events.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The change LGTM, just minor nitpicks
@@ -332,6 +332,9 @@ TEST_F(DependsOnTests, ShortcutFunctionWithWaitList) { | |||
ASSERT_NE(Cmd, nullptr); | |||
Cmd->MIsBlockable = true; | |||
Cmd->MEnqueueStatus = detail::EnqueueResultT::SyclEnqueueBlocked; | |||
// for the test functionality, depenent task HostTaskEvent must be treated as |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I would replace this comment with "Mock up an incomplete host task" on L326.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Done.
@intel/llvm-reviewers-runtime Colleagues, could you please look at the PR? |
sycl/source/detail/queue_impl.cpp
Outdated
} | ||
|
||
if (isInOrder()) { | ||
auto &EventToStoreIn = MGraph.expired() ? MDefaultGraphDeps.LastEventPtr | ||
: MExtGraphDeps.LastEventPtr; | ||
EventToStoreIn = EventImpl; | ||
EventToStoreIn = std::move(EventImpl); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Could we make EventImpl
a const auto &
instead? That way, there would be no point in moving it here and we would avoid conditionally invalid future attempts at accessing EventImpl
after this point.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Done.
1) Aling scheduler bypass conditions for CG and for memory ops. 2) Connect an event returned from scheduler bypass case with dependent events.
Co-authored-by: Sergey Semenov <[email protected]>
03df21e
to
a4fa35a
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks! I think it looks good, but I also would like to give @sergey-semenov another day to look.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM