#001awesome-kernel-skills1 skills242atualizado 2026-07-07100% do criadorskillocupaçãodescriçãoatualizadocutedsl-kernelsDesenvolvedores de softwareWrite, debug, validate, and optimize NVIDIA CuTe DSL (nvidia-cutlass-dsl) GPU kernels in Python. Use for CuTe layout algebra, TV layouts, tiled copies, predication, shared/register/tensor memory, cp.async and TMA pipelines, mbarriers, warp specialization, warp/block/cluster reductions, MMA on SM80 (tensor cores), SM90 (WGMMA), and SM100/Blackwell (tcgen05/UMMA, TMEM, block-scaled FP4/FP8), tile schedulers, torch interop via from_dlpack, PTX inspection, and diagnosing alignment/vectorization/synchronization failures.2026-07-07Mostrando as 1 principais de 1 skills coletadas neste repositório.Carregar mais 0 skillsCarregando skills...