Setting up #ComfyUI for Linux with Radeon GPU ended being way easier than anticipated.
1. Install latest version of ROCm.
2. With pip, install a pyTorch build for that ROCm version.
3. Clone ComfyUI repo.
4. Create a Python virtual environment and activate it.
5. Install the Python packages from requirements.
Then if everything was done right, you'll be able to bring up ComfyUI server without errors. The interface can be reached via browser from localhost port 8188.
Aethernal McAwesome
npub1rt…02pj2
2026-07-05 01:56:39 UTC
Author Public Key
npub1rttwjkwzjtm5mes46nrwtmzr6zm7cjgg54w7jw4z2f7yd29ar4dsk02pj2Published at
2026-07-05 01:56:39 UTCEvent JSON
{
"id": "5e24bc96ecc64c7cd2eab02f233be667f5a67b03f1569b423c76cdb02d3f18f7",
"pubkey": "1ad6e959c292f74de615d4c6e5ec43d0b7ec4908a55de93aa2527c46a8bd1d5b",
"created_at": 1783216599,
"kind": 1,
"tags": [
[
"alt",
"A short note: Setting up #ComfyUI for Linux with Radeon GPU ende..."
],
[
"t",
"ComfyUI"
],
[
"t",
"comfyui"
]
],
"content": "Setting up #ComfyUI for Linux with Radeon GPU ended being way easier than anticipated.\n\n1. Install latest version of ROCm.\n2. With pip, install a pyTorch build for that ROCm version. \n3. Clone ComfyUI repo.\n4. Create a Python virtual environment and activate it.\n5. Install the Python packages from requirements.\n\nThen if everything was done right, you'll be able to bring up ComfyUI server without errors. The interface can be reached via browser from localhost port 8188.",
"sig": "a541ea48fc03d43d748057a9b35ed03e00a0e2ae87f02274bf8570feee4baf3e77a44af8ef37e9a9eb5b232573a3331bd1713ed941eb4fba6cb4e6eb5f960378"
}