High-Performance llama.cpp fork for Legacy GPUs: Kepler Support, CUDA Graphs & Windows NCCL!

🌐 链接: https://www.reddit.com/r/LocalLLaMA/comments/1xeicw2/highperformance_llamacpp_fork_for_legacy_gpus/
🏷️ 分组: Reddit｜AI·自托管·VPS
🕒 时间: 2026-10-25 04:01:46
Reddit
From the LocalLLaMA community on Reddit: High-Performance llama.cpp fork for Legacy GPUs: Kepler Support, CUDA Graphs & Windows…
From the LocalLLaMA community on Reddit: High-Performance llama.cpp fork for Legacy GPUs: Kepler Support, CUDA Graphs & Windows…
Explore this post and more from the LocalLLaMA community