Last released Jun 27, 2026
In-kernel %globaltimer profiler for warp-specialized CUDA kernels (Perfetto/Chrome timelines).
Last released Mar 23, 2026
A general x-bit quantization engine for LLMs,[2-8] bits, awq/gptq/hqq/vptq
Last released Jan 3, 2026
A multi-backend SYRK (Symmetric Rank-K Update) implementation
Supported by