Support running different allocation analyses #2605
Merged
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
Parametrize `Allocation::run` to support more runnning more allocation analyses, thus making the allocation analysis framework more easily extensible. `ModuleAllocation` instances are now obtained from a templated `get` static member function to propagate the analysis kind to use. Signed-off-by: victor-eds <[email protected]>
victor-eds
commented
Oct 31, 2024
Comment on lines
-253
to
+263
| explicit ModuleAllocation(ModuleOp moduleOp) | ||
| : CallGraph<Allocation>(moduleOp) { | ||
| walk<WalkOrder::PreOrder, WalkOrder::PostOrder>( | ||
| template <typename AllocationAnalysis = triton::AllocationAnalysis> | ||
| static ModuleAllocation get(ModuleOp moduleOp) { | ||
| ModuleAllocation res(moduleOp); | ||
| res.walk<WalkOrder::PreOrder, WalkOrder::PostOrder>( |
Contributor
Author
There was a problem hiding this comment.
Cannot have templated constructors like this, so this is a static member function now.
…ysis` This is just a copy of `mlir::triton::AllocationAnalysis` for now. Signed-off-by: victor-eds <[email protected]>
victor-eds
force-pushed
the
specialize-allocation-analysis
branch
from
d3fe27b to
68bf33c
Compare
victor-eds
commented
Oct 31, 2024
Contributor
Author
There was a problem hiding this comment.
Do not review analysis implementation. This is a copy of Allocation.cpp.
chengjunlu
reviewed
Nov 1, 2024
Contributor
|
Does this PR solves the issue about the allocation #2340 |
Contributor
Author
This is a first PR refactoring the code so we can modify freely and fix issues like that one and the ones I'm finding with my current FA work |
etiotto
reviewed
Nov 1, 2024
Contributor
Author
|
|
chengjunlu
approved these changes
Nov 4, 2024
etiotto
approved these changes
Nov 4, 2024
sommerlukas
approved these changes
Nov 5, 2024
Contributor
Author
|
Just rebased and solved conflicts. Enabling auto-merge. |
victor-eds
added a commit
to victor-eds/intel-xpu-backend-for-triton
that referenced
this pull request
Nov 26, 2024
intel#2605 introduced allocation analysis specialization downstream, but upstreaming lead to a different approach (triton-lang/triton#5070). This commit reverts residual changes to the analysis matching upstream's triton-lang/triton@deee78f. Signed-off-by: victor-eds <[email protected]>
etiotto
pushed a commit
that referenced
this pull request
Nov 26, 2024
#2605 introduced allocation analysis specialization downstream, but upstreaming lead to a different approach (triton-lang/triton#5070). This commit reverts residual changes to the analysis matching upstream's triton-lang/triton@deee78f. Closes #2825. Signed-off-by: victor-eds <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Each of the two commits in this PR should be reviewed separately.
08786ac: To be upstreamed. Extends memory allocation analysis infrastructure so the analysis to use can be configured.
d3fe27b: Downstream. Use an Intel-specific analysis. This is just a copy of the original for now, so, please, do not review the analysis implementation.
96a2e78 and cbfdaf9 should be part of the first commit. Will squash there when upstreaming.