This post is a graphical overview of the WebGPU x Go open-source landscape as of 2026, comparing four projects for high-performance graphics and GPU computing in Go.
Key points
- gogpu/wgpu (Recommended · Pure Go Implementation)
- Pure Go implementation: no Rust dependency, no CGO
- Multi-backend support: Vulkan, Metal, DX12, GLES, Software
- Fully self-controlled, simple compilation, easy cross-platform deployment
- go-webgpu (Active · Zero-CGO Bindings)
- Built on wgpu-native
- Uses goffi bindings, no C compiler required
- Complete API coverage, currently in Beta stage
- cogentcore (Versatile · Cross Platform)
- Supports Web/Wasm
- iOS/Android support
- Feature-rich with well-developed examples
- rajveermalviya (Archived · Native Bindings)
- Early pioneering project in the Go WebGPU space
- Served as the foundation for CogentCore
- Now archived and read-only
Conclusion
The post argues that combining WebGPU's modern GPU API with Go's efficient development experience opens possibilities for game development, scientific computing, and data visualization. The pure-Go gogpu/wgpu is highlighted as the recommended option, while go-webgpu (bindings-based) and cogentcore (cross-platform) serve different use cases, and the original rajveermalviya bindings remain available as a historical reference.