[rc/3.2.x] LLVM bump for gfx950 target support#5417
Merged
Conversation
This reverts commit 7e401df.
This reverts commit 2d8093c.
Fixes triton-lang#4769 (cherry picked from commit f484cb8)
triton-lang#5064) Bumping llvm to include a loop unroller fix: llvm/llvm-project#114573. This is needed for subsequent loop unroller upstreaming work. (cherry picked from commit 3c296ab)
This pulls in llvm/llvm-project@bd9145c8c213 to enable ASan on AMD backend. (cherry picked from commit 0bd30a2)
This includes llvm/llvm-project#115627 (cherry picked from commit 6404fbb)
This relands triton-lang#5392 to enable new arch target since backend support has been added--it doesn't depend on the reverted LLVM upgrade in triton-lang#5341; basic necessary enablement is already included in the current llvm version we're using. (cherry picked from commit f257479)
jataylo
requested review from
antiagainst,
ptillet and
zhanglx13
as code owners
jataylo
changed the title
Contributor
|
@jataylo , it seems like the dpp implementation for reduce is reverted, is that intentional? Thanks |
Contributor
Author
Intentional, we initially were going to bring this in as a cherry pick to the release branch but we're simplifying things to just focus on gfx950 support as the release branch is already cut. |
Collaborator
|
cc @bertmaher could you please review this PR ? |
Contributor
Author
|
cc: @bertmaher, @atalman Note this PR looks a bit more complicated than it actually is because we are doing reverts for some of the optimizations we brought into rc/3.2.x earlier to make things cleaner for the release/3.2.x cherry pick |
atalman
approved these changes
Dec 13, 2024
Collaborator
atalman
left a comment
atalman
left a comment
There was a problem hiding this comment.
lgtm, approving to be able to test in pytorch nightly
pruthvistony
approved these changes
Dec 13, 2024
jataylo
added a commit
to jataylo/triton
that referenced
this pull request
Dec 18, 2024
This PR brings in required LLVM bumps and additional targets for gfx950 support. - triton-lang#5040 - triton-lang#5064 - triton-lang#5180 - triton-lang#5242 - triton-lang#5392 Note this PR reverts the last two PRs to only focus on the LLVM upgrade - triton-lang#5347 - triton-lang#5191 --------- Co-authored-by: peterbell10 <peterbell10@openai.com> Co-authored-by: Hongtao Yu <hoy@meta.com> Co-authored-by: Lei Zhang <antiagainst@gmail.com> Co-authored-by: Jungwook Park <jungwook.park@amd.com> (cherry picked from commit f11c5ba)
jataylo
added a commit
to jataylo/triton
that referenced
this pull request
Dec 18, 2024
This PR brings in required LLVM bumps and additional targets for gfx950 support. - triton-lang#5040 - triton-lang#5064 - triton-lang#5180 - triton-lang#5242 - triton-lang#5392 Note this PR reverts the last two PRs to only focus on the LLVM upgrade - triton-lang#5347 - triton-lang#5191 --------- Co-authored-by: peterbell10 <peterbell10@openai.com> Co-authored-by: Hongtao Yu <hoy@meta.com> Co-authored-by: Lei Zhang <antiagainst@gmail.com> Co-authored-by: Jungwook Park <jungwook.park@amd.com> (cherry picked from commit f11c5ba)
bertmaher
pushed a commit
that referenced
this pull request
Dec 19, 2024
This PR brings in required LLVM bumps and additional targets for gfx950 support. - #5040 - #5064 - #5180 - #5242 - #5392 Note this PR reverts the last two PRs to only focus on the LLVM upgrade - #5347 - #5191 --------- Co-authored-by: peterbell10 <peterbell10@openai.com> Co-authored-by: Hongtao Yu <hoy@meta.com> Co-authored-by: Lei Zhang <antiagainst@gmail.com> Co-authored-by: Jungwook Park <jungwook.park@amd.com>
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
8 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.
This PR brings in required LLVM bumps and additional targets for gfx950 support.
Note this PR reverts the last two PRs to only focus on the LLVM upgrade