Agent Skills by Fan
Showing 4 vetted skills indexed across 1 GitHub repositories.
fmh6614
kernel-benchmark
Benchmark custom GPU kernels against PyTorch baselines with CUDA event timing.
Community
Advanced
fmh6614
kernel-profile
Profile CUDA-C++ kernels with Nsight Compute metrics and classify bottlenecks.
Community
Intermediate
fmh6614
kernel-loop
Orchestrate evidence-gated one-variable optimization loops for CUDA, CUTLASS, CuTe DSL, or Triton GPU kernels.
Community
Advanced
fmh6614
kernel-KBS
Retrieve evidence-backed GPU kernel documentation from a SQLite corpus.
Community
Intermediate