-
Notifications
You must be signed in to change notification settings - Fork 673
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Multiple issues installing for AMD GPU (Radeon RX7600XT) #1519
Comments
Thanks for reporting. You're right that the Unfortunately the preview branch is in a broken state at the moment on ROCm. We're working on it! The commit at |
Thanks, I am now using fork by AMD (https://github.com/ROCm/bitsandbytes) which seems to be working; what is advise? use that or commit at |
I think either should be fine but if you've already achieved a working build then I would personally stick to that. |
I had exactly the same issue:
Reverting to commit
This seems to relate to issue #1492 . Is this correct? Anything else I can do to make it work in the meantime? |
Also checked the bitsandbytes AMD fork but this resulted in the same issue I'm using the latest PyTorch nightly:
|
triton-lang/triton#5471 |
I assume you mean |
Yeah. |
FYI:
|
@Lier0 @mcondarelli Did you mean Edit: I now see I have
But that results in the issue How to get |
@Lier0 I figured it out, thanks to @mcondarelli script. The correct command is:
In your reply you didn't add the |
Works for me. Thank you! |
Thank you for bringing this up and the discussion. We will try to understand the issues here and get back to you. |
The error still holds true with latest wheel installed as: pip install https://github.com/bitsandbytes-foundation/bitsandbytes/releases/download/continuous-release_multi-backend-refactor/bitsandbytes-1.0.0-py3-none-manylinux_2_24_x86_64.whl Please fix before releasing v1.0.0.
|
System Info
I am under Linux Mint Xia (based on Ubuntu 24.04).
CPU: AMD Ryzen 9 5950X 16-Core Processor with 64GiB RAM.
GPU: Advanced Micro Devices, Inc. [AMD/ATI] Navi 33 [Radeon RX 7600/7600 XT/7600M XT/7600S/7700S / PRO W7600] (rev c0) (Actually: RX 7600 XT, if it matters)
Python: Python 3.10.16
Application: I am testing with
InvokeAI
Reproduction
I tried following recipe but I found several errors:
.toml
and thuspip install -r requirements-dev.txt
won't work.cmake -DCOMPUTE_BACKEND=hip -S . && make
completes with no errors (just a few "kernels.hip:2857:17: warning: loop not unrolled:...".I also tried:
Expected behavior
I expected to be able to use
bitsandbytes
.The text was updated successfully, but these errors were encountered: