ashvardanian/FatWheels
Starter kit for fat Python wheels: CPU kernels multi-versioned for AVX2, SVE & RVV, plus CUDA & ROCm kernels from one source, validated under QEMU, with reusable Clang, QEMU, GenMC, CUDA & ROCm CI images 🛞
GitHub repository with 31 stars and 4 forks.
Language: C++
Topics: arm-sve, avx2, cpython, cuda, docker, fat-binary, hip, multiversioning, openmp, python