WgMath
Open GPU scientific computing for every platform
Today’s GPU scientific computing ecosystem is still strongly dominated by CUDA, a closed, proprietary technology tied to a specific hardware vendor. The WgMath project aims to empower the scientific computing community, including the web community, with a collection of foundational GPU mathematical libraries that are fully cross-platform (hence not tied to a specific hardware vendor) by leveraging the open WebGPU standard, as well as WebAssembly for browser support. WgMath will provide mathematical compute shaders for linear algebra, geometry, and rigid-body physics simulation; as well as some utilities for easily combining WGSL shaders through Rust libraries and its popular Cargo dependencies management tool. With the creation of these foundational libraries, we aim to promote the development of a scientific computing community building highly performant, reusable, cross-platform, scientific computing projects, while relying on open standards, and preserving freedom of GPU hardware selection.
- The project's own website: https://dimforge.com
Run by Dimforge Eurl
This project was funded through the NGI0 Core Fund, a fund established by NLnet with financial support from the European Commission's Next Generation Internet programme, under the aegis of DG Communications Networks, Content and Technology under grant agreement No 101092990.
