This repository provides an incredibly easy way to compare NumPy performance across the following builds on your machine:
- official PyPI NumPy wheel pypi
- conda NumPy builds (OpenBLAS / MKL) numpy libblas
- official AOCL wheel amd
curl -fsSL https://pixi.sh/install.sh | sh
pixi install -a
./benchmark.sh
pixi run -e pypi dot
pixi run -e openblas dot
pixi run -e mkl dot
pixi run -e aocl dot
pixi run -e blis dot
pixi run -e netlib dot
pixi list -e mkl | grep libblas
Environment: mkl
libblas 3.11.0 4_h5875eb1_mkl 18.4 KiB conda https://conda.anaconda.org/conda-forge/
pixi list -e openblas | grep libblas
Environment: openblas
libblas 3.11.0 4_h4a7cf45_openblas 18.1 KiB conda https://conda.anaconda.org/conda-forge/
pixi run -e pypi doctor
pixi run -e openblas doctor
pixi run -e mkl doctor
pixi run -e aocl doctor
This distribution includes third-party components. See LICENSES/ for the corresponding licenses.