amd-apu

AMD APU compatible Ollama. Get up and running with OpenAI gpt-oss, DeepSeek-R1, Gemma 3 and other models.

Latest Bazzite Deck, GNOME, and KDE images for AMD BC-250 boards with cyan-skillfish-governor-smu, GPU frequency scaling, MangoHud/radeontop 655% telemetry fix, signed rebases, and optional experimental 40CU runtime images

Measured inference stack for Strix Halo (Ryzen AI Max+ 395, 128 GB) — llama.cpp serving plus TTS/TTI/TTV media workloads under one memory authority. Not a fork: master plus a short list of patches. Memory budget, defect registry, prefix-cache gateway.

Unified memory ML inference engine for AMD APU (RDNA 3.5 / gfx1150) — Custom HIP kernels with 2.2x speedup over standard allocation

🚀 Automated nightly builds & multi-backend releases (ROCm 7 TheRock, Vulkan RADV, CUDA, Metal) for CachyLLama & llama-ai on AMD Strix Halo (8060S), Strix Point (890M), Phoenix (780M), and Steam Deck with Homebrew support.

APU-Codec: Neural audio codec from source, optimized for AMD APU tri-processor inference (NPU encoder, GPU decoder)

Custom AI Web Studio & ComfyUI backend optimized for AMD ROCm 7.14, RDNA 3.5 APUs (Radeon 890M), FLUX.1 GGUF, and MiniMax Music generation.

⚡ Automated nightly builds & portable ROCm 7 releases of Ember for DeepSeek-V4-Flash on AMD Strix Halo (gfx1151 / Radeon 8060S). Standalone C inference server with bundled ROCm runtime, continuous batching, and Homebrew support.

SUPERFAST: turn an AMD Ryzen AI Max+ 395 (Strix Halo, gfx1151) into a fast, private local-LLM machine. Four profiles at maximum context: Qwen3.8-27B dense, Qwen3.8-Flash-Next MoE, Gemma-4-26B, DeepSeek-V4-Flash. Fedora 44 setup script, verified downloads, one-command switch, agent-ready API.