Wasi: WebGPU – A Proposed WebAssembly System Interface API (github.com)

A new proposal, wasi:webgpu, has been introduced to create a WebAssembly System Interface (WASI) API designed for GPU access in WebAssembly environments. This initiative aims to leverage the inherent benefits of WebAssembly, such as portability, security, and sandboxing, while extending GPU compute capabilities to a wider range of applications. Use cases highlighted for this API include server-side graphics streaming, scientific computing, AI/ML inference and training, image and video processing, and data visualization. While the API is based on the official WebGPU specification, it includes notable deviations to better suit non-browser environments, ensuring clarity on these adjustments is part of the documentation. The significance of wasi:webgpu lies in its potential to empower diverse industries, particularly AI and machine learning, by enabling efficient GPU utilization across various operating systems including Linux, Windows, macOS, and Android. The integration of GPU compute resources into the WebAssembly ecosystem could revolutionize how intensive calculations are performed, making advanced graphics and AI applications more accessible and performance-driven without the usual constraints of traditional web environments. As the proposal moves towards its next phase, stakeholders are encouraged to provide feedback and engage in discussions, further refining this promising technology.