saymrwulf/onnxruntime
1
0
Fork 0
mirror of https://github.com/saymrwulf/onnxruntime.git synced 2026-07-26 19:52:38 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

[WebNN EP] Add WebNN doc (#20733)

Browse source 
- Add WebNN EP build instruction
- Update WebNN Chromium flag in tutorial
- Some minor updates
This commit is contained in:
Wanming Lin 2024-05-21 15:17:31 +08:00 committed by GitHub
parent 35cd15e6a9
commit 944db3ed6c
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 30 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

33
docs/build/web.md vendored
View file

@ -70,7 +70,7 @@ This support is added/removed by appending the following flags to the build comm
| `--enable_wasm_threads` | build with multi-thread support |
| `--enable_wasm_simd` | build with SIMD support |
ONNX Runtime Web can be built with WebGPU support via JavaScript Execution Provider (JSEP). To build with JSEP support, use flag `--use_jsep`.
ONNX Runtime Web can be built with WebGPU and WebNN support via JavaScript Execution Provider (JSEP). To build with JSEP support, use flag `--use_jsep`. Building WebNN support requires additional flag `--use_webnn`.
ONNX Runtime Web can also be built to support the training APIs. To build with training APIs included, use the flag `--enable-training-apis`.
@ -95,22 +95,23 @@ in `<ORT_ROOT>/`, run one of the following commands to build WebAssembly:
A full list of required build artifacts:
| file name | file name (renamed) | build flag used |
| --------------------------- | -------------------------------- | --------------------------------------------------------- |
| ort-wasm.js | | |
| ort-wasm.wasm | | |
| ort-wasm-threaded.js | | `--enable_wasm_threads` |
| ort-wasm-threaded.wasm | | `--enable_wasm_threads` |
| ort-wasm-threaded.worker.js | | `--enable_wasm_threads` |
| ort-wasm-simd.wasm | | `--enable_wasm_simd` |
| ort-wasm-simd-threaded.wasm | | `--enable_wasm_simd` `--enable_wasm_threads` |
| ort-wasm-simd.js | ort-wasm-simd.jsep.js | `--use_jsep` `--enable_wasm_simd` |
| ort-wasm-simd.wasm | ort-wasm-simd.jsep.wasm | `--use_jsep` `--enable_wasm_simd` |
| ort-wasm-simd-threaded.js | ort-wasm-simd-threaded.jsep.js | `--use_jsep` `--enable_wasm_simd` `--enable_wasm_threads` |
| ort-wasm-simd-threaded.wasm | ort-wasm-simd-threaded.jsep.wasm | `--use_jsep` `--enable_wasm_simd` `--enable_wasm_threads` |
| ort-training-wasm-simd.wasm | | `--enable_wasm_simd` `--enable_training_apis` |
| file name | file name (renamed) | build flag used |
| --------------------------- | -------------------------------- | ----------------------------------------------------------------------- |
| ort-wasm.js | | |
| ort-wasm.wasm | | |
| ort-wasm-threaded.js | | `--enable_wasm_threads` |
| ort-wasm-threaded.wasm | | `--enable_wasm_threads` |
| ort-wasm-threaded.worker.js | | `--enable_wasm_threads` |
| ort-wasm-simd.wasm | | `--enable_wasm_simd` |
| ort-wasm-simd-threaded.wasm | | `--enable_wasm_simd` `--enable_wasm_threads` |
| ort-wasm-simd.js | ort-wasm-simd.jsep.js | `--use_jsep` `--use_webnn` `--enable_wasm_simd` |
| ort-wasm-simd.wasm | ort-wasm-simd.jsep.wasm | `--use_jsep` `--use_webnn` `--enable_wasm_simd` |
| ort-wasm-simd-threaded.js | ort-wasm-simd-threaded.jsep.js | `--use_jsep` `--use_webnn` `--enable_wasm_simd` `--enable_wasm_threads` |
| ort-wasm-simd-threaded.wasm | ort-wasm-simd-threaded.jsep.wasm | `--use_jsep` `--use_webnn` `--enable_wasm_simd` `--enable_wasm_threads` |
| ort-training-wasm-simd.wasm | | `--enable_wasm_simd` `--enable_training_apis` |
NOTE: WebGPU and WebNN is currently supported as experimental feature for ONNX Runtime Web. The build instructions may change. Please make sure to refer to latest documents from [webgpu gist](https://gist.github.com/fs-eire/a55b2c7e10a6864b9602c279b8b75dce) and [webnn gist](https://gist.github.com/Honry/88b87c43b3f51a6c38c10454f3599405) for a detailed build/consume instruction for ORT Web WebGPU and WebNN.
NOTE: WebGPU is currently supported as experimental feature for ONNX Runtime Web. The build instructions may change. Please make sure to refer to latest documents from [this gist](https://gist.github.com/fs-eire/a55b2c7e10a6864b9602c279b8b75dce) for a detailed build/consume instruction for ORT Web WebGpu.
### Minimal Build Support

13
docs/get-started/with-javascript/web.md
View file

@ -50,6 +50,17 @@ import * as ort from 'onnxruntime-web/webgpu';
const ort = require('onnxruntime-web/webgpu');
```
If you want to use ONNX Runtime Web with WebNN support (experimental feature), you need to import as below:
```js
// use ES6 style import syntax (recommended)
import * as ort from 'onnxruntime-web/experimental';
```
```js
// or use CommonJS style import syntax
const ort = require('onnxruntime-web/experimental');
```
For a complete table for importing, see [Conditional Importing](https://github.com/microsoft/onnxruntime-inference-examples/tree/main/js/importing_onnxruntime-web#conditional-importing).
## Documentation
@ -98,4 +109,4 @@ The following are video tutorials that use ONNX Runtime Web in web applications:
- \[2]: WebGPU requires Chromium v113 or later on Windows. Float16 support requires Chrome v121 or later, and Edge v122 or later.
- \[3]: WebGPU requires Chromium v121 or later on Windows.
- \[4]: WebGL support is in maintenance mode. It is recommended to use WebGPU for better performance.
- \[5]: Requires to launch browser with commandline flag `--enable-experimental-web-platform-features`.
- \[5]: Requires to launch browser with commandline flag `--enable-features=WebMachineLearningNeuralNetwork`.

2
docs/tutorials/web/index.md
View file

@ -32,7 +32,7 @@ For more detail on the steps below, see the [build a web application](./build-we
You can also use the onnxruntime-web package in the frontend of an electron app.
With onnxruntime-web, you have the option to use `webgl` or `webgpu` for GPU processing, and WebAssembly (`wasm`, alias to `cpu`) for CPU processing. All ONNX operators are supported by WASM but only a subset are currently supported by WebGL and WebGPU.
With onnxruntime-web, you have the option to use `webgl`, `webgpu` or `webnn` (with `deviceType` set to `gpu`) for GPU processing, and WebAssembly (`wasm`, alias to `cpu`) or `webnn` (with `deviceType` set to `cpu`) for CPU processing. All ONNX operators are supported by WASM but only a subset are currently supported by WebGL, WebGPU and WebNN.
* Inference on server in JavaScript. Use the `onnxruntime-node` package.