microsoft / microsoft/onnxruntime
docs: add missing documentation for SNPE, VSINPU, WebGPU, and WebNN execution providers
- Dominant language
- C++
- Stars
- 21.9k
- Forks
- 4.2k
- Avg merge
- 4d 11h
- Merged PRs (30d)
- 184
Description
### Description
The following execution providers have code implementations in the ONNX Runtime repository but lack documentation pages on the [onnxruntime.ai](https://onnxruntime.ai/docs/execution-providers/) documentation site:
| Execution Provider | Code Location | Current Docs |
|---|---|---|
| **SNPE** | `onnxruntime/core/providers/snpe/` | None |
| **VSINPU** | `onnxruntime/core/providers/vsinpu/` | None |
| **WebGPU** | `onnxruntime/core/providers/webgpu/` | Tutorial only (`docs/tutorials/web/ep-webgpu.md`) — no dedicated EP page |
| **WebNN** | `onnxruntime/core/providers/webnn/` | Tutorial only (`docs/tutorials/web/ep-webnn.md`) — no dedicated EP page |
### Request
Add proper EP documentation pages (on the `gh-pages` branch under `docs/execution-providers/`) for each of these providers, following the existing EP page format which typically includes:
- Overview and supported platforms
- Build instructions / prerequisites
- Configuration options and provider-specific settings
- Supported operators / known limitations
- Usage examples
### Context
This was discovered during a documentation audit comparing all EP code directories against the gh-pages documentation. WebGPU and WebNN are particularly high-impact since they are actively used web platform EPs with only tutorial-level docs.
For SNPE and VSINPU, input from the respective EP maintainers would be needed to write accurate documentation.
Contributor guide
Research direction
Start by reviewing the existing pages under docs/execution-providers/ on the gh-pages branch, then compare them with the SNPE, VSINPU, WebGPU, and WebNN code directories and the existing WebGPU/WebNN tutorials. Done means four provider pages covering the requested sections, with maintainer input for SNPE and VSINPU where needed.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- documentation, machine-learning
- Issue type
- Documentation
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 48/100