nvbench
nvbench copied to clipboard
Create Python bindings to nvbench
Using Python API to nvbench one could
- use Python as benchmark driver
- analyze benchmark data in Python as they are collected
- profile kernels authored in Python, such as
- numba kernels
- warp kernels,
- kernels generated by CuPy, torch, etc.
- kernels hand-written using
cuda.core
- use
nvbenchas backend for other Python benchmarking projects