[SYCL][Graph] Memory reuse for graph allocations in a single graph #18340
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
Contributor
Bensuo
commented
May 6, 2025
Bensuo
commented
- Optimize memory use for allocations within a single graph by reusing memory where possible
- New handler impl member for node dependency access with the CGF
- New E2E tests for memory reuse
- Add missing CGType -> string conversion for graph printing alloc and free nodes
Bensuo
force-pushed
the
ben/graph-memory-reuse
branch
from
dec8b75 to
cdcec7e
Compare
Bensuo
force-pushed
the
ben/graph-memory-reuse
branch
from
cdcec7e to
64fef78
Compare
- Optimize memory use for allocations within a single graph by reusing memory where possible - New handler impl member for node dependency access with the CGF - New E2E tests for memory reuse - Add design doc information on memory reuse - Add missing CGType -> string conversion for graph printing alloc and free nodes
Bensuo
force-pushed
the
ben/graph-memory-reuse
branch
from
64fef78 to
c1d9595
Compare
EwanC
reviewed
May 8, 2025
sycl/test-e2e/Graph/AsyncAlloc/Inputs/async_alloc_device_memory_reuse.cpp
Show resolved
Hide resolved
EwanC
approved these changes
May 8, 2025
fabiomestre
reviewed
May 9, 2025
- Fix wording in design doc - Remove code duplication is async_alloc calls - Early return from tryReuseExistingAllocation()
- No dep nodes means that we cannot be connected to any free nodes for reuse
Contributor
Author
|
@intel/llvm-reviewers-runtime Could I get a review on this please? The Arc failures on CI look unrelated to this PR. |
againull
approved these changes
May 19, 2025
Contributor
|
@Bensuo @EwanC https://github.com/intel/llvm/actions/runs/15128095339/job/42523770858 To me it looks like a bogus warning but I'm not an expert in this code. Can someone take a look and work around the warning if it is bogus? Thanks |
Contributor
Author
I think this is bogus yeah, there is other similar initialization in the runtime that isn't getting flagged on GCC 11.4, and the error doesn't show up in 12.3. I've reworked the code so it's not initialized outside and then assigned to in the lambda (and fixed a few small issues I saw at the same time) and that seems fine now: #18583 |
Contributor
|
Thanks! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
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.