[ROCm] Add pieces from the CUDA codgen lowering path #14769
Merged
Conversation
This file contains 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
Enable iree-llvmgpu-cast-address-space-function. Add DropSharedMemoryDeallocOp rewrite pattern in ConvertToROCDLPass. Remove dynamic legalization for FuncOp. Add ConvertToDynamicSharedMemory ConvertToROCDLPass.
MaheshRavishankar
requested changes
Aug 21, 2023
| // A `dealloc` is converted into a call to `free` on the underlying data buffer. | ||
| // The memref descriptor being an SSA value, there is no need to clean it up | ||
| // in any way. | ||
| struct DropSharedMemoryDeallocOp : public OpRewritePattern<memref::DeallocOp> { |
There was a problem hiding this comment.
THe norm is to not have the pattern exposed directly this way. Could you move these patterns to Common/GPU and use a populateDropSharedMemoryDeallocPatterns method to add these patterns to both paths?
|
@MaheshRavishankar, will you rereview this PR? |
MaheshRavishankar
approved these changes
Aug 28, 2023
|
@sogartar is this ready to merge? |
It is. |
jinchen62
pushed a commit
to jinchen62/iree
that referenced
this pull request
Sep 18, 2023
Enable iree-llvmgpu-cast-address-space-function. Add DropSharedMemoryDeallocOp rewrite pattern in ConvertToROCDLPass. Remove dynamic legalization for FuncOp. Add ConvertToDynamicSharedMemory ConvertToROCDLPass. I am not at all confident that these need to be the same on the ROCm path as are on the CUDA path. If there is no one to express confidence I will dig deeper to make sure it makes sense. --------- Co-authored-by: Boian Petkantchin <boian@nod-labs.com>
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Outdated suggestions cannot be applied.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.
Enable iree-llvmgpu-cast-address-space-function.
Add DropSharedMemoryDeallocOp rewrite pattern in ConvertToROCDLPass. Remove dynamic legalization for FuncOp.
Add ConvertToDynamicSharedMemory ConvertToROCDLPass.
I am not at all confident that these need to be the same on the ROCm path as are on the CUDA path. If there is no one to express confidence I will dig deeper to make sure it makes sense.