AMD Hipfire - a new inference engine optimized for AMD GPU's
Hipfire, a new inference engine optimized for AMD GPUs, uses mq4 quantization and shows significant speedups on Localmaxxing benchmark.
··
Archive
The Archive
Search the full wire by company, model, lab, or keyword. Every story we have ever aggregated.
Thoughts on using an AMD Alveo V80 FPGA PCI card as a poor man’s Taalas HC1 (LLM-burned-onto-a-chip).
Community exploration of AMD Alveo V80 FPGA as alternative to Taalas HC1 LLM accelerator, targeting 1.4–3.2k tokens/sec with quantized models.
··
Comparison of upcoming x86 unified memory systems
Roadmap comparison of x86 unified-memory APUs: AMD Gorgon Halo (2024), Intel Nova Lake AX (2025), AMD Medusa Halo (2027) with 50% perf gains.
··
Qwen3.6 35B-A3B is quite useful on 780m iGPU (llama.cpp,vulkan)
Qwen 3.6 35B-A3B MoE model achieves 250+ tok/s on AMD Radeon 780M iGPU via llama.cpp Vulkan.
··
AMD and OpenAI announce strategic partnership to deploy 6 gigawatts of AMD GPUs
AMD and OpenAI commit to deploying 6 gigawatts of AMD Instinct GPUs starting 2026.
·
Ollama now supports AMD graphics cards
Ollama now supports AMD graphics cards in preview on Windows and Linux. All the features of Ollama can now be accelerated by AMD graphics cards on Ollama for Linux and Windows.
·
← Front Page13 matches