Run the CUDA kernels on ROCm - #10
Conversation
Signed-off-by: Kashif Rasul <kashif.rasul@gmail.com>
|
All contributors have signed the CLA ✍️ ✅ |
|
I have read the CLA Document and I hereby sign the CLA |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 1bea03137b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Signed-off-by: Kashif Rasul <kashif.rasul@gmail.com>
Signed-off-by: Kashif Rasul <kashif.rasul@gmail.com>
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 79c171f989
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Signed-off-by: Kashif Rasul <kashif.rasul@gmail.com>
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: d28502180f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
…fy cache Signed-off-by: Kashif Rasul <kashif.rasul@gmail.com>
1 participant
This makes the compiled flashrnn kernels build and run on AMD GPUs through HIP, while leaving the CUDA path unchanged.
What's in here:
Tested on ROCm 7.2 (gfx1150, torch 2.13+rocm7.2): the full tests/test_slstm.py suite passes, 870 tests, covering the cuda and cuda_fused backends for both forward and backward.