gpu-io

A library for running GPU-accelerated computations, such as physics simulations, in a web browser using WebGL. It helps developers do heavy parallel calculations without a server.

Share on XLicense: MIT

Overview

gpu-io is a WebGL library for building GPU-accelerated computations that run in a web browser. It handles WebGL state, shader and program caching, and differences between WebGL versions and browsers, so you write less boilerplate while keeping low-level control. It targets WebGL 2 with a fallback to WebGL 1, and WebGPU support is planned. Typical uses include physics and particle simulations, cellular automata, image processing and general GPU calculations, with rendering straight to a canvas.

Key features

  • Manages WebGL state and caches shaders and programs
  • WebGL 2 with fallback to WebGL 1
  • Renders directly to the WebGL canvas
  • Suits simulations, cellular automata and image processing

Best for

Good for web developers who want GPU-based simulations or image work without writing raw WebGL. WebGPU support is only planned, not available yet.

Upstream
amandaghassaei/gpu-io
Fork on GitHub
Guo-astro/gpu-io
Upstream stars
1.5k
Category
Developer tools and infrastructure
License
MIT
Forked
2025-10-17
Sync status
In syncLast synced 2026-09-29