[SYCL] Pass handler & instead of queue across ABI for reduction utils
#18834
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
aelovikov-intel
had a problem deploying
to
WindowsCILock
GitHub Actions
Error
|
This is based on top of #18830, please ignore first commit during review. Also, while @intel/sycl-graphs-reviewers aren't technically owners of the reduction-related changes, it mostly affects the graph-related code path, so their review is still welcome. |
aelovikov-intel
force-pushed
the
handler-in-red-utils
branch
from
86d2491 to
442a4c2
Compare
aelovikov-intel
had a problem deploying
to
WindowsCILock
GitHub Actions
Failure
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
aelovikov-intel
force-pushed
the
handler-in-red-utils
branch
from
442a4c2 to
6f81b65
Compare
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
Queue might be `nullptr` in case of graph, but the information this utils query is device-specific. By passing entire `handler &` and having access to graph information we'd be able to return more precise results. Another positive side-effect is that we eliminiate explicit `std::shared_ptr<queue_impl>` which is a small step forward in the ongoing refactoring efforts to prefer passing `*_impl` by raw ptr/ref with explicit `shared_from_this` whenever lifetimes need to be extended.
aelovikov-intel
force-pushed
the
handler-in-red-utils
branch
from
6f81b65 to
a580702
Compare
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
Rebased, not relevant anymore. @intel/sycl-graphs-reviewers review request still stands :) @slawekptak, this is ready for review. |
aelovikov-intel
added a commit
to aelovikov-intel/llvm
that referenced
this pull request
Jun 6, 2025
Initially started in intel#18830 Subsequent PRs before this final one: intel#18794 intel#18834
fabiomestre
approved these changes
Jun 6, 2025
There was a problem hiding this comment.
I think this is fine since graphs at the moment don't support reductions: https://github.com/intel/llvm/blob/sycl/sycl/doc/extensions/experimental/sycl_ext_oneapi_graph.asciidoc#12-non-implemented-features-and-known-issues
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
uditagarwal97
approved these changes
Jun 6, 2025
aelovikov-intel
added a commit
to aelovikov-intel/llvm
that referenced
this pull request
Jun 6, 2025
Initially started in intel#18830 Subsequent PRs before this final one: intel#18794 intel#18834
aelovikov-intel
added a commit
to aelovikov-intel/llvm
that referenced
this pull request
Jun 6, 2025
Initially started in intel#18830 Subsequent PRs before this final one: intel#18794 intel#18834 intel#18748
aelovikov-intel
added a commit
to aelovikov-intel/llvm
that referenced
this pull request
Jun 6, 2025
Initially started in intel#18830 Subsequent PRs before this final one: intel#18794 intel#18834 intel#18748
aelovikov-intel
added a commit
to aelovikov-intel/llvm
that referenced
this pull request
Jun 6, 2025
Initially started in intel#18830 Subsequent PRs before this final one: intel#18794 intel#18834 intel#18748
aelovikov-intel
added a commit
to aelovikov-intel/llvm
that referenced
this pull request
Jun 6, 2025
Initially started in intel#18830 Subsequent PRs before this final one: intel#18794 intel#18834 intel#18748
# for free
to join this conversation on GitHub.
Already have an account?
# 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.
Queue might be
nullptrin case of graph, but the information theseutilitiess query is device-specific. By passing entire
handler &and havingaccess to graph information we'd be able to return more precise results.
Another positive side-effect is that we eliminiate explicit
std::shared_ptr<queue_impl>which is a small step forward in theongoing refactoring efforts to prefer passing
*_implby raw ptr/refwith explicit
shared_from_thiswhenever lifetimes need to be extended.