Nathanw1014 / strix-halo-llamacpp
Performance-tuned llama.cpp for AMD Strix Halo (gfx1151): FA + MoE-prefill fixes with a bundled current Mesa driver. Vulkan and HIP; portable dir, Docker, and distrobox.
Why this job: says: llama.cpp; topics: llama-cpp, local-llm, llm-inference
What it does
A ready-to-run llama.cpp for AMD Strix Halo (Ryzen AI Max+ 395 / Radeon 8060S / gfx1151), tuned for long-context, quantized-KV workloads. It bundles a set of Flash-Attention and MoE-prefill fixes plus a current GPU driver, so quantized KV cache is fast instead of a penalty.
Get started
No single install command. Read the steps in the README (opens in new tab)
- Maintenance
- Active
- Last commit 16 days ago
- Stars
- 224
- +2 this week
- Licence
- Custom
- Read before business use
- Downloads for
- None listed
- No release files
- Latest version
- v0.8.0
- Released 16 days ago
- Built with
- Python
- 59% of the code
Is it looked after?
- Contributors
- Not stated
- Last 90 days
- Issues
- 83% closed
- 4 still open
- Pull requests
- 0 merged
- 0 waiting
- Releases
- 78
- 30+ in the last year
- No contributing guide
- No security policy
- No code of conduct
Similar repos
| Repo | Stars | Last commit | Licence | Downloads for |
|---|---|---|---|---|
| hogeheer499-commits / strix-halo-guide | 371 | yesterday | MIT | No downloads |
| peonist-ai / halogen-flash-server | 1,275 | yesterday | Custom | No downloads |
| shawnshekari / strixite | 100 | 5 days ago | AGPL-3.0 | No downloads |
More: languages, topics and project record
Languages
- Python 59%
- Shell 41%
- HIP 1%
Topics
- amd
- flash-attention
- gfx1151
- llama-cpp
- llm-inference
- local-llm
- radv
- rocm
- ryzen-ai-max
- speculative-decoding
- strix-halo
- vulkan
Project record
- JobLocal LLMs
- StatusListed
- Commits104
- Forks21
- Watching7
- Stars in 28 days+9
- CreatedJul 26, 2026
- First seen by usNov 7, 2026
- CheckedNov 8, 2026