Skip to content

Document AMD ROCm install for Windows in README #1559

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.

Sign up for GitHub

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

Draft
dxqb wants to merge 1 commit into
base: master
Choose a base branch
from
Draft

Document AMD ROCm install for Windows in README #1559

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 20 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,6 +65,26 @@ OneTrainer is a one-stop solution for all your Diffusion training needs.
>
> Additionally it's been reported Alpine, Arch and Xubuntu Linux may be missing `tkinter`. Install it via `apk add py3-tk` for Alpine and `sudo pacman -S tk` for Arch.

#### AMD (ROCm)

**Linux:** works out of the box, run `install.sh` as normal.

**Windows:** manual install:

1. Find your GPU's GFX architecture from [AMD's table](https://github.com/ROCm/TheRock/blob/main/RELEASES.md#supported-python-device--install-extras) (e.g. `gfx1100`).
2. Create and activate a venv, then (replace `gfx####`):

@dxqb dxqb Jun 28, 2026

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

give the commands. not all users know how to do that

@NeuralFault NeuralFault Jun 28, 2026 •
edited
Loading

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It should be the same procedure as described in the section on LN52-54. Can just refer to Step 1 thru 4 above

```
pip install torch[device-gfx####] torchvision[device-gfx####] --index-url https://repo.amd.com/rocm/whl-multi-arch/
pip install -r requirements-global.txt onnxruntime triton-windows
```
3. Set environment variables:
- `MIOPEN_FIND_MODE=2`
- `FLASH_ATTENTION_TRITON_AMD_ENABLE=TRUE`
- `PYTORCH_ALLOC_CONF=max_split_size_mb:512,garbage_collection_threshold:0.8`
- `TORCH_ROCM_AOTRITON_ENABLE_EXPERIMENTAL=1` (RDNA 3 / RX 7000 or newer)
- `HIP_VISIBLE_DEVICES=1` (only if you also have an AMD iGPU)
4. *(Optional)* bitsandbytes (Python 3.12 only): install [this wheel](https://github.com/0xDELUXA/bitsandbytes_win_rocm/releases), then set `ROCM_PATH` and `HIP_PATH` when launching.
Comment thread
dxqb marked this conversation as resolved.

## Updating

#### Automatic update
Expand Down