Issues

• SM86 GGUF Linear: cooperative loads and a direct BF16 MMQ epilogue

  iamwavecut/ninfer-all#107 · Filter by author liqlos opened on Oct 19, 2026
  1

• [Feature Request] Add optional CPU input embedding offload to free VRAM

  iamwavecut/ninfer-all#104 · Filter by author witherhoard99 opened on Oct 19, 2026

• RTX 4090 field report: NInfer-all vs llama.cpp for long-context Qwen coding (+ thanks!)

  iamwavecut/ninfer-all#96 · Filter by author spk-c opened on Oct 17, 2026
  1

• Is anyone already working on NVFP4 support for pre-Blackwell GPUs?

  iamwavecut/ninfer-all#86 · Filter by author 13thblade opened on Oct 16, 2026

• Runs really good, and fast!

  iamwavecut/ninfer-all#70 · Filter by author jabbor opened on Oct 13, 2026

• Feature request: one running instance with both Qwen chat and Jev-style decision capability

  iamwavecut/ninfer-all#64 · Filter by author lededev opened on Oct 13, 2026
  2