Integrate llvm-project@faf5d747f174cc #20828
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
Integrate torch-mlir@389541fb9ddd Integrate stablehlo@5837b2a6ce - Drop revert of the 1:N dialect conversion removal since all dependencies have migrated - Drop revert of the APIntParameter error since all dependencies have migrated - Drop revert of allowing function type conversion to fail since all dependencies have migrated - Drop local modifications to torch-mlir and stablehlo, those are now clean submodules (they're no longer needed now that we've dropped reverts and they've migrated) - We still have a revert of upstream #137930 since it's not clear the roccertness issue is resolved (that's the SDWA cndmask thing, see llvm/llvm-project#138766 ) - We still have a revert of upstream #133231 since that could still be breaking tests - Rename bufferization.to_memref to bufferization.to_buffer everewhere - Swap all getSource() on Transfer*Op to getBase() - Rename the memref argument of TransferGatherOp to `base` to match the transfer interface - Remove argument materialization calls - In the one case where this wasn't trivial, migrate to a target materilazition since that's what upstream advice suggested - Handle (in any way that seemed appropriate) the failures that eraseArguments() and eraseResults() can now have
benvanik
approved these changes
May 15, 2025
|
Ok, that did turn out to be complicated. Thanks @krzysz00 |
|
thanks @krzysz00! |
|
(CC @Muzammiluddin-Syed-ECE who is on integrate duty this week.) |
|
Many thanks, @krzysz00 ! |
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Outdated suggestions cannot be applied.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.
Integrate torch-mlir@389541fb9ddd
Integrate stablehlo@5837b2a6ce
Dropped reverts
Continued reverts
Changes
baseto match the transfer interfaceaffine-expand-index-opsafter the lastFoldMemRefAliasOpscall, becauseFoldMemRefAliasOpsnow createsaffine.linearize_indexandaffine.delinearize_indexwhich don't seem to lower to LLVM right on their own. See [mlir][MemRef] Use specialized index ops to fold expand/collapse_shape llvm/llvm-project#138930