Running MiniMax H3 on AMD GPUs: full ComfyUI + ROCm 7.14 Ubuntu guide
Apprehensive_Sky892 · reddit · 2026-09-13
A Reddit user published a hands-on tutorial for running MiniMax H3 (MMH3) video generation in ComfyUI on AMD GPUs — RDNA3 (RX 7900 series) and RDNA4 (RX 9070, AI Pro R9700) — under Ubuntu.
Key configuration:
- Requires ROCm 7.14.0; look up your GPU's gfx???? codename in AMD's spec table and install matching torch 2.12.0 / torchvision 0.27.0 / torchaudio 2.11.0 from AMD's official wheel index.
- Why Linux over Windows: H3 is heavy, so you want all VRAM and RAM for generation; AMD on Linux is also more stable and faster.
- Hardware requirement: RDNA3/4 card with 16GB+ VRAM.
- Detailed steps cover making an Ubuntu Server (minimized) boot drive with RUFUS (MBR, UEFI, Secure Boot off), installing to an external USB drive to spare the main disk, partitioning tricks (500M EFI, odd-size partitions like 42G to avoid wiping the wrong drive), and netplan/UEFI caveats for cross-machine installs.
A complete, follow-along guide for running heavyweight video models locally on the AMD side.
Related event: Guide: running MiniMax H3 on AMD GPUs with ComfyUI and ROCm(2 posts)→
More from Infra
- Google DeepMind team publishes deep dive on scaling Transformer inference — zainhas · 2026-09-13
- Open weights need orders-of-magnitude cheaper training or it's a LARP, researcher argues — jd_pressman · 2026-09-13
- Local intelligence movement calls for builders as the case for edge compute strengthens weekly — VoidStateKate · 2026-09-13
- Chip embargo backfired: GLM and Kimi now served at scale on Huawei chips, argues Toulme — sudoraohacker · 2026-09-13
- Running ComfyUI+MMH3 on AMD GPUs: RX 9070, AI Pro R9700 and RX 7900 tested — Apprehensive_Sky892 · 2026-09-13
- Why AI infra engineers hate model dynamism: a full-screen scream — generativist · 2026-09-13