Replies: 18 comments 76 replies
-
|
This is great! I'll have a closer look in the coming days and help you get this landed and in the ci. |
Beta Was this translation helpful? Give feedback.
-
|
Awesome ! I’ll try this tonight |
Beta Was this translation helpful? Give feedback.
-
|
@jammm, compared to the mainline-snapshot-2025-03-13 tarball and current main branch, my build contains the following changes:
|
Beta Was this translation helpful? Give feedback.
-
|
I've pushed my container images to Dockerhub at https://hub.docker.com/r/scottt/therock/tags and editing the post to reflect that. |
Beta Was this translation helpful? Give feedback.
-
|
I've pushed some Dockerfile failed to build from a clean slate fixes previous masked by my placing the source trees in a cache mount.
|
Beta Was this translation helpful? Give feedback.
-
|
@scottt I tried your container but the result of gpt2 script always ends up giving:
Oddly enough, if I run the script with
And running something simple like |
Beta Was this translation helpful? Give feedback.
-
|
I'v found that MIOpen built with current main branch options on gfx1151 would not work with
I have two patches that appears to successfully workaround the problem:
With these changes, I can now run AUTOMATIC1111/stable-diffusion-webui |
Beta Was this translation helpful? Give feedback.
-
|
This image has my MIOpen fixes: https://hub.docker.com/layers/scottt/therock/pytorch-vision-dev-f41/images/sha256-3e80ce26fa02bf4a566ed693e1d6741cf8aaa742319f688e7c584d5a83bb47c8 |
Beta Was this translation helpful? Give feedback.
-
|
Couple TODOs:
|
Beta Was this translation helpful? Give feedback.
-
|
I got some massive perf improvements after compiling aotriton for gfx1151 so it uses the flash-attention/memory-efficient backends when running scaled_dot_product_attention: before: after: above benchmark script: Using a couple real-world models: |
Beta Was this translation helpful? Give feedback.
-
|
Continuing the results from #244 (reply in thread), Input image (from the sd3.5 output on gfx1151) Output video (dolly-zoom-out camera trajectory, had to downscale the gif to be under the 10mb attachment limit) mp4: Just had to make sure pytorch was compiled with There's one caveat though - multiple times the GPU would crash out with segfaults like |
Beta Was this translation helpful? Give feedback.
-
|
@stellaraccident , for the first part of the gfx1151 enablement work, could you take a look at merging #357 ? I bumped the hipBLASLt submodule to include the gfx1151 support from upstream but had to update the patches in TheRock for how hipBLASLt finds tools, which was moderately painful. |
Beta Was this translation helpful? Give feedback.
-
|
@stellaraccident , could you take at merging #392 for the MIOpen portion of the gfx1151 work? |
Beta Was this translation helpful? Give feedback.
-
|
@stellaraccident , @jammm , I've filed the PR for gfx1151 Pytorch work here: #449 |
Beta Was this translation helpful? Give feedback.
-
|
FYI, we've identified a root cause for ROCm/MIOpen#3685 but don't have a fix yet |
Beta Was this translation helpful? Give feedback.
-
|
While I am running the command below for sudo docker buildx build --build-arg AMDGPU_TARGETS=gfx1151 --file dockerfiles/pytorch-dev/pytorch_dev_ubuntu_24.04.Dockerfile .I got the error below: Let me know if there is a patch for that. |
Beta Was this translation helpful? Give feedback.
-
|
any docs to integrate this to get it working with vllm? |
Beta Was this translation helpful? Give feedback.
-
|
i installed rocm therock nightly tarball
therock-dist-linux-gfx1151-7.0.0rc20250714.tar.gz and rocm pytorch from
https://d2awnip2yjpvqn.cloudfront.net/v2/gfx1151/
and tried to compile https://github.com/ROCm/vllm/ repo but build failed.
when amd has added the gfx1151 support i assume they will provide the
docker image on https://hub.docker.com/r/rocm/vllm-dev
…On Fri, Aug 15, 2025 at 3:21 PM medioxor ***@***.***> wrote:
@scottt <https://github.com/scottt> @bugbuster-dev
<https://github.com/bugbuster-dev> any luck with getting vllm working?
—
Reply to this email directly, view it on GitHub
<#244 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ARD6KQCSAJ3KFPUKEIXR2ML3NXNEXAVCNFSM6AAAAABZSYJAR6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTIMJRGY3TENQ>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.




Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
I see others like https://github.com/jammm are also working on gfx1151
I have pytorch, pytorch-vision, and hipBLASLt working well enough to run GPT2 on the Strix Halo in an Asus Z13 here and thought others might want to take a look https://github.com/scottt/rocm-TheRock/commits/gfx1151/
(I'm running Bazzite Linux and the distro kernel)
To download the image from Dockerhub:
Alternatively, to build the images from source:
Running GPT2 through
huggingface/transformersthen appears to work:With the content of
gpt2.pybeing:Beta Was this translation helpful? Give feedback.
All reactions