ViperEkura 28d11f1610 chore: relocate kernel benchmarks to csrc/bench ...

- Move benchmark_gemv.py, benchmark_swiglu.py, and benchmark_gemv_common.py from scripts/tools/ to csrc/bench/ so kernel benchmarks live next to the kernels they measure
- Update reproduction commands in decode_linear_benchmark.md, swiglu_benchmark.md, and cuda_kernels.md
- Codify the placement convention in AGENTS.md: kernel benchmarks in csrc/bench/, pure-CUDA harnesses in csrc/tests/*.cu, engine and evaluation benchmarks in scripts/

2026-09-03 21:06:33 +08:00

..

bench

chore: relocate kernel benchmarks to csrc/bench

2026-09-03 21:06:33 +08:00

kernels

fix: resolve audited dispatch, kernel, and rollout bugs

2026-09-03 16:58:14 +08:00

tests

refactor: root kernel includes at csrc/kernels

2026-09-01 14:43:38 +08:00

__init__.py

refactor: reorganize CUDA kernels into per-family directories

2026-08-22 20:40:31 +08:00

CMakeLists.txt

refactor: drop gemv variant shape tables and flatten kernel dir

2026-09-03 06:34:20 +08:00