Reference implementations of fundamental GPU programming patterns with MXMACA.

A high perfromance op kernel lib running on metax HD platform

A kernel library written in tilelang

Python bindings to the MetaX Management Library(MXSML)

Go bindings for the MetaX System Management Library(MXSML)

High-performance LLM operator library built on TileLang.

Open deep learning compiler stack for cpu, gpu and specialized accelerators

Fast and efficient attention method exploration and implementation.

Community maintained hardware plugin for vLLM on MetaX GPU