onnxruntime/js/web/lib/wasm/proxy-wrapper.ts

220 lines
7.6 KiB
TypeScript
Raw Normal View History

// Copyright (c) Microsoft Corporation. All rights reserved.
// Licensed under the MIT License.
import {env, InferenceSession} from 'onnxruntime-common';
import {OrtWasmMessage, SerializableInternalBuffer, SerializableSessionMetadata, SerializableTensorMetadata, TensorMetadata} from './proxy-messages';
import * as core from './wasm-core-impl';
import {initializeWebAssembly} from './wasm-factory';
const isProxy = (): boolean => !!env.wasm.proxy && typeof document !== 'undefined';
let proxyWorker: Worker|undefined;
let initializing = false;
let initialized = false;
let aborted = false;
type PromiseCallbacks<T = void> = [resolve: (result: T) => void, reject: (reason: unknown) => void];
let initWasmCallbacks: PromiseCallbacks;
const queuedCallbacks: Map<OrtWasmMessage['type'], Array<PromiseCallbacks<unknown>>> = new Map();
const enqueueCallbacks = (type: OrtWasmMessage['type'], callbacks: PromiseCallbacks<unknown>): void => {
const queue = queuedCallbacks.get(type);
if (queue) {
queue.push(callbacks);
} else {
queuedCallbacks.set(type, [callbacks]);
}
};
const ensureWorker = (): void => {
if (initializing || !initialized || aborted || !proxyWorker) {
throw new Error('worker not ready');
}
};
const onProxyWorkerMessage = (ev: MessageEvent<OrtWasmMessage>): void => {
switch (ev.data.type) {
case 'init-wasm':
initializing = false;
if (ev.data.err) {
aborted = true;
initWasmCallbacks[1](ev.data.err);
} else {
initialized = true;
initWasmCallbacks[0]();
}
break;
case 'init-ep':
case 'copy-from':
case 'create':
case 'release':
case 'run':
case 'end-profiling': {
const callbacks = queuedCallbacks.get(ev.data.type)!;
if (ev.data.err) {
callbacks.shift()![1](ev.data.err);
} else {
callbacks.shift()![0](ev.data.out!);
}
break;
}
default:
}
};
const scriptSrc = typeof document !== 'undefined' ? (document?.currentScript as HTMLScriptElement)?.src : undefined;
export const initializeWebAssemblyAndOrtRuntime = async(): Promise<void> => {
if (initialized) {
return;
}
if (initializing) {
throw new Error('multiple calls to \'initWasm()\' detected.');
}
if (aborted) {
throw new Error('previous call to \'initWasm()\' failed.');
}
initializing = true;
if (!BUILD_DEFS.DISABLE_WASM_PROXY && isProxy()) {
// overwrite wasm filepaths
if (env.wasm.wasmPaths === undefined) {
if (scriptSrc && scriptSrc.indexOf('blob:') !== 0) {
env.wasm.wasmPaths = scriptSrc.substr(0, +(scriptSrc).lastIndexOf('/') + 1);
}
}
return new Promise<void>((resolve, reject) => {
proxyWorker?.terminate();
const workerUrl = URL.createObjectURL(new Blob(
[
// This require() function is handled by esbuild plugin to load file content as string.
// eslint-disable-next-line @typescript-eslint/no-require-imports
require('./proxy-worker/main')
],
{type: 'text/javascript'}));
proxyWorker = new Worker(workerUrl, {name: 'ort-wasm-proxy-worker'});
proxyWorker.onerror = (ev: ErrorEvent) => reject(ev);
proxyWorker.onmessage = onProxyWorkerMessage;
URL.revokeObjectURL(workerUrl);
initWasmCallbacks = [resolve, reject];
const message: OrtWasmMessage = {type: 'init-wasm', in : env};
proxyWorker.postMessage(message);
});
} else {
try {
await initializeWebAssembly(env.wasm);
await core.initRuntime(env);
initialized = true;
} catch (e) {
aborted = true;
throw e;
} finally {
initializing = false;
}
}
};
export const initializeOrtEp = async(epName: string): Promise<void> => {
if (!BUILD_DEFS.DISABLE_WASM_PROXY && isProxy()) {
ensureWorker();
return new Promise<void>((resolve, reject) => {
enqueueCallbacks('init-ep', [resolve, reject]);
const message: OrtWasmMessage = {type: 'init-ep', in : {epName, env}};
proxyWorker!.postMessage(message);
});
} else {
await core.initEp(env, epName);
}
};
export const copyFromExternalBuffer = async(buffer: Uint8Array): Promise<SerializableInternalBuffer> => {
if (!BUILD_DEFS.DISABLE_WASM_PROXY && isProxy()) {
ensureWorker();
return new Promise<SerializableInternalBuffer>((resolve, reject) => {
enqueueCallbacks('copy-from', [resolve, reject]);
const message: OrtWasmMessage = {type: 'copy-from', in : {buffer}};
proxyWorker!.postMessage(message, [buffer.buffer]);
});
} else {
return core.copyFromExternalBuffer(buffer);
}
};
export const createSession =
async(model: SerializableInternalBuffer|Uint8Array, options?: InferenceSession.SessionOptions):
Promise<SerializableSessionMetadata> => {
if (!BUILD_DEFS.DISABLE_WASM_PROXY && isProxy()) {
// check unsupported options
if (options?.preferredOutputLocation) {
throw new Error('session option "preferredOutputLocation" is not supported for proxy.');
}
ensureWorker();
return new Promise<SerializableSessionMetadata>((resolve, reject) => {
enqueueCallbacks('create', [resolve, reject]);
[js/web] rewrite backend resolve to allow multiple EPs (#19735) ### Description This PR rewrite the backend resolve logic to support specifying multiple EPs. #### Backend The first version of ONNX Runtime Web actually carried some existing code from [ONNX.js](https://github.com/microsoft/onnxjs), which includes the "backend" concept. The original "backend" in ONNX.js is designed in a way assuming there is only one backend from user's backend hint list will be used. For example, in ONNX.js, if user specify a backend hint as `['webgl', 'wasm']`, ONNX.js will first try to use WebGL backend - if it loads successfully (the browser supports webgl), then "webgl" backend will be used and "wasm" will be ignored; otherwise, "webgl" will be ignored and try to load "wasm" backend. In short: only one backend will be used when initializing a session. #### Execution Provider Execution Provider, or EP, in ONNX Runtime is a different concept. One of the differences is that users are allow to specify multiple EPs, and if one does not support a particular kernel, it can fallback to other EP. This is a very common case when using a GPU EP in ONNX Runtime. #### Current Status: Backend v.s. EP Because of the history reasons mentioned above, the current status is quite confusing. There are **real backend**s, which means it's different implementation in code; and there are **backend hint**s, which are used as string names for backend hint; and there are **EP**s of the ONNX Runtime concepts. currently there are only 2 **backend**s in our code base: The "onnxjs backend", and the "wasm backend". The "onnxjs backend" currently only powers backend hint "webgl", which go into the old onnx.js code path. All other backend hints including "wasm", "cpu"(alias to wasm), "webgpu" and "webnn" are all powered by "wasm backend". And because ORT Web treat "backend" as an internal concept and want to align with ONNX Runtime, so those names of backend hints are becoming EP names. The following table shows today's status: | Execution Provider Name (public) / Backend Hint (internal) | Backend | EP in ORT | -------- | ------- | ------- | | "wasm"/"cpu" | WasmBackend | CPU EP | "webgl" | OnnxjsBackend | \* technically not an EP | "webgpu" | WasmBackend | JSEP | "webnn" | WasmBackend | WebNN EP #### Problem While the API allows to specify multiple EPs, the backend resolving only allows one backend. This causes issues when user specify multiple EP names in session options, the backend resolve behavior and EP registration behavior is inconsistent. Specifically, in this issue: https://github.com/microsoft/onnxruntime/issues/15796#issuecomment-1925363908: EP list `['webgpu', 'wasm']` on a browser without WebGPU support resolves to 'wasm' backend, but the full EP list is passed in session options, so JSEP is still enabled, causing the runtime error. #### Solution Since we still need WebGL backend, we cannot totally remove the backend register/resolve system. In this PR I made the following changes: - initialize every backend from the EP list, instead of only do that for the first successful one. - for the first resolved backend, filter all EP using the exact same backend. Remove all EPs not using this backend from session options - for every explicitly specified EP, if it's removed, show a warning message in console
2024-03-15 18:47:45 +00:00
const message: OrtWasmMessage = {type: 'create', in : {model, options: {...options}}};
const transferable: Transferable[] = [];
if (model instanceof Uint8Array) {
transferable.push(model.buffer);
}
proxyWorker!.postMessage(message, transferable);
});
} else {
return core.createSession(model, options);
}
};
export const releaseSession = async(sessionId: number): Promise<void> => {
if (!BUILD_DEFS.DISABLE_WASM_PROXY && isProxy()) {
ensureWorker();
return new Promise<void>((resolve, reject) => {
enqueueCallbacks('release', [resolve, reject]);
const message: OrtWasmMessage = {type: 'release', in : sessionId};
proxyWorker!.postMessage(message);
});
} else {
core.releaseSession(sessionId);
}
};
export const run = async(
[js/webgpu] support IO binding (#17480) <del> **This PR is based on a few prerequisites PRs. They are listed as below:** - #17465 - #17469 - #17470 - #17472 - #17473 - #17484 Please review the current change by only looking at commit e2e6623e673ec6de55a5c1f8edcbd3a46b535a89 and later. </del> ### Description This PR introduces WebGPU IO binding. This new feature allows onnxruntime-web users to use tensors created from GPU as model input/output so that a model inferencing can be done without unnecessary data copy between CPU and GPU for model input/output. ### Examples An E2E demo/example is being worked on. Following is some simple demo with code snippet. Let's first check today how we do: ```js // STEP.1 - create an inference session: const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'] }); // STEP.2 - create model input: (supposing myImageCpuData is a Float32Array) const feeds = { 'input_image:0': new ort.Tensor('float32', myImageCpuData, [1, 224, 224, 3]) }; // STEP.3 - run model const myResults = await mySession.run(feeds); // STEP.4 - get output data const myData = myResults['output_image:0'].data; // Float32Array ``` #### for inputs (GPU tensor): Now, with IO binding, you can create a tensor from a GPU buffer, and feed it to the model: ```js // new STEP.2.A - create model input from a GPU buffer: (supposing myInputGpuBuffer is a `GPUBuffer` object with input data) const feeds = { 'input_image:0': ort.Tensor.fromGpuBuffer(myInputGpuBuffer, { dataType: 'float32', dims: [1, 224, 224, 3] }) }; ``` ### for outputs (pre-allocated GPU tensor) you can also do that for output, **if you know the output shape**: ```js // new STEP.2.B - create model output from a GPU buffer: (supposing myOutputGpuBuffer is a pre-allocated `GPUBuffer` object) const fetches = { 'output_image:0': ort.Tensor.fromGpuBuffer(myOutputGpuBuffer, { dataType: 'float32', dims: [1, 512, 512, 3] }) }; // new STEP.3 - run model with pre-allocated output (fetches) const myResults = await mySession.run(feeds, fetches); ``` ### for outputs (specify location) if you do not know the output shape, you can specify the output location when creating the session: ```js // new STEP.1 - create an inference session with an option "preferredOutputLocation": const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'], preferredOutputLocation: "gpu-buffer" }); ``` if the model has multiple outputs, you can specify them seperately: ```js // new STEP.1 - create an inference session with an option "preferredOutputLocation": const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'], preferredOutputLocation: { "output_image:0": "gpu-buffer" } }); ``` now you don't need to prepare the `fetches` object and onnxruntime-web will prepare output data on the location that specified. #### read data when you get the output tensor, you can: ```js // get the gpu buffer object: const gpuBuffer = myOutputTensor.gpuBuffer; // GPUBuffer // get the CPU data asynchronizely const cpuData = await myOutputTensor.getData(); // get the CPU data asynchronizely and release the underlying GPU resources const cpuData = await myOutputTensor.getData(true); // dispose the tensor (release the underlying GPU resources). This tensor object will be invalid after dispose() is called. myOutputTensor.dispose(); ``` #### resource management JavaScript has GC so you don't need to worry about managing JavaScript objects. But there are 2 types of resources that are not managed by GC: - GPU buffer that used in tensors - Underlying ORT native resources To simplify, most of the unmanaged resources and handled inside ORT web. But there are a few resources that need users to manage: - All external GPU resources, including GPU buffers inside all tensors created by `Tensor.fromGpuBuffer()`, will not be managed by ORT. User should manage those GPU buffers themselves. - When a session is created with `preferredOutputLocation` == "gpu-buffer" specified in session options, and the corresponding output is not pre-allocated, user need to call the output tensor's `dispose()` or `getData(true)` to manually release the underlying GPU buffers. - ORT internal errors (including providing a pre-allocated output tensor with wrong type/dims) will invalidate the whole wasm memory and is not recoverable. An exception is thrown in this situation.
2023-09-29 18:24:42 +00:00
sessionId: number, inputIndices: number[], inputs: TensorMetadata[], outputIndices: number[],
outputs: Array<TensorMetadata|null>, options: InferenceSession.RunOptions): Promise<TensorMetadata[]> => {
if (!BUILD_DEFS.DISABLE_WASM_PROXY && isProxy()) {
[js/webgpu] support IO binding (#17480) <del> **This PR is based on a few prerequisites PRs. They are listed as below:** - #17465 - #17469 - #17470 - #17472 - #17473 - #17484 Please review the current change by only looking at commit e2e6623e673ec6de55a5c1f8edcbd3a46b535a89 and later. </del> ### Description This PR introduces WebGPU IO binding. This new feature allows onnxruntime-web users to use tensors created from GPU as model input/output so that a model inferencing can be done without unnecessary data copy between CPU and GPU for model input/output. ### Examples An E2E demo/example is being worked on. Following is some simple demo with code snippet. Let's first check today how we do: ```js // STEP.1 - create an inference session: const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'] }); // STEP.2 - create model input: (supposing myImageCpuData is a Float32Array) const feeds = { 'input_image:0': new ort.Tensor('float32', myImageCpuData, [1, 224, 224, 3]) }; // STEP.3 - run model const myResults = await mySession.run(feeds); // STEP.4 - get output data const myData = myResults['output_image:0'].data; // Float32Array ``` #### for inputs (GPU tensor): Now, with IO binding, you can create a tensor from a GPU buffer, and feed it to the model: ```js // new STEP.2.A - create model input from a GPU buffer: (supposing myInputGpuBuffer is a `GPUBuffer` object with input data) const feeds = { 'input_image:0': ort.Tensor.fromGpuBuffer(myInputGpuBuffer, { dataType: 'float32', dims: [1, 224, 224, 3] }) }; ``` ### for outputs (pre-allocated GPU tensor) you can also do that for output, **if you know the output shape**: ```js // new STEP.2.B - create model output from a GPU buffer: (supposing myOutputGpuBuffer is a pre-allocated `GPUBuffer` object) const fetches = { 'output_image:0': ort.Tensor.fromGpuBuffer(myOutputGpuBuffer, { dataType: 'float32', dims: [1, 512, 512, 3] }) }; // new STEP.3 - run model with pre-allocated output (fetches) const myResults = await mySession.run(feeds, fetches); ``` ### for outputs (specify location) if you do not know the output shape, you can specify the output location when creating the session: ```js // new STEP.1 - create an inference session with an option "preferredOutputLocation": const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'], preferredOutputLocation: "gpu-buffer" }); ``` if the model has multiple outputs, you can specify them seperately: ```js // new STEP.1 - create an inference session with an option "preferredOutputLocation": const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'], preferredOutputLocation: { "output_image:0": "gpu-buffer" } }); ``` now you don't need to prepare the `fetches` object and onnxruntime-web will prepare output data on the location that specified. #### read data when you get the output tensor, you can: ```js // get the gpu buffer object: const gpuBuffer = myOutputTensor.gpuBuffer; // GPUBuffer // get the CPU data asynchronizely const cpuData = await myOutputTensor.getData(); // get the CPU data asynchronizely and release the underlying GPU resources const cpuData = await myOutputTensor.getData(true); // dispose the tensor (release the underlying GPU resources). This tensor object will be invalid after dispose() is called. myOutputTensor.dispose(); ``` #### resource management JavaScript has GC so you don't need to worry about managing JavaScript objects. But there are 2 types of resources that are not managed by GC: - GPU buffer that used in tensors - Underlying ORT native resources To simplify, most of the unmanaged resources and handled inside ORT web. But there are a few resources that need users to manage: - All external GPU resources, including GPU buffers inside all tensors created by `Tensor.fromGpuBuffer()`, will not be managed by ORT. User should manage those GPU buffers themselves. - When a session is created with `preferredOutputLocation` == "gpu-buffer" specified in session options, and the corresponding output is not pre-allocated, user need to call the output tensor's `dispose()` or `getData(true)` to manually release the underlying GPU buffers. - ORT internal errors (including providing a pre-allocated output tensor with wrong type/dims) will invalidate the whole wasm memory and is not recoverable. An exception is thrown in this situation.
2023-09-29 18:24:42 +00:00
// check inputs location
if (inputs.some(t => t[3] !== 'cpu')) {
throw new Error('input tensor on GPU is not supported for proxy.');
}
// check outputs location
if (outputs.some(t => t)) {
throw new Error('pre-allocated output tensor is not supported for proxy.');
}
ensureWorker();
[js/webgpu] support IO binding (#17480) <del> **This PR is based on a few prerequisites PRs. They are listed as below:** - #17465 - #17469 - #17470 - #17472 - #17473 - #17484 Please review the current change by only looking at commit e2e6623e673ec6de55a5c1f8edcbd3a46b535a89 and later. </del> ### Description This PR introduces WebGPU IO binding. This new feature allows onnxruntime-web users to use tensors created from GPU as model input/output so that a model inferencing can be done without unnecessary data copy between CPU and GPU for model input/output. ### Examples An E2E demo/example is being worked on. Following is some simple demo with code snippet. Let's first check today how we do: ```js // STEP.1 - create an inference session: const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'] }); // STEP.2 - create model input: (supposing myImageCpuData is a Float32Array) const feeds = { 'input_image:0': new ort.Tensor('float32', myImageCpuData, [1, 224, 224, 3]) }; // STEP.3 - run model const myResults = await mySession.run(feeds); // STEP.4 - get output data const myData = myResults['output_image:0'].data; // Float32Array ``` #### for inputs (GPU tensor): Now, with IO binding, you can create a tensor from a GPU buffer, and feed it to the model: ```js // new STEP.2.A - create model input from a GPU buffer: (supposing myInputGpuBuffer is a `GPUBuffer` object with input data) const feeds = { 'input_image:0': ort.Tensor.fromGpuBuffer(myInputGpuBuffer, { dataType: 'float32', dims: [1, 224, 224, 3] }) }; ``` ### for outputs (pre-allocated GPU tensor) you can also do that for output, **if you know the output shape**: ```js // new STEP.2.B - create model output from a GPU buffer: (supposing myOutputGpuBuffer is a pre-allocated `GPUBuffer` object) const fetches = { 'output_image:0': ort.Tensor.fromGpuBuffer(myOutputGpuBuffer, { dataType: 'float32', dims: [1, 512, 512, 3] }) }; // new STEP.3 - run model with pre-allocated output (fetches) const myResults = await mySession.run(feeds, fetches); ``` ### for outputs (specify location) if you do not know the output shape, you can specify the output location when creating the session: ```js // new STEP.1 - create an inference session with an option "preferredOutputLocation": const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'], preferredOutputLocation: "gpu-buffer" }); ``` if the model has multiple outputs, you can specify them seperately: ```js // new STEP.1 - create an inference session with an option "preferredOutputLocation": const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'], preferredOutputLocation: { "output_image:0": "gpu-buffer" } }); ``` now you don't need to prepare the `fetches` object and onnxruntime-web will prepare output data on the location that specified. #### read data when you get the output tensor, you can: ```js // get the gpu buffer object: const gpuBuffer = myOutputTensor.gpuBuffer; // GPUBuffer // get the CPU data asynchronizely const cpuData = await myOutputTensor.getData(); // get the CPU data asynchronizely and release the underlying GPU resources const cpuData = await myOutputTensor.getData(true); // dispose the tensor (release the underlying GPU resources). This tensor object will be invalid after dispose() is called. myOutputTensor.dispose(); ``` #### resource management JavaScript has GC so you don't need to worry about managing JavaScript objects. But there are 2 types of resources that are not managed by GC: - GPU buffer that used in tensors - Underlying ORT native resources To simplify, most of the unmanaged resources and handled inside ORT web. But there are a few resources that need users to manage: - All external GPU resources, including GPU buffers inside all tensors created by `Tensor.fromGpuBuffer()`, will not be managed by ORT. User should manage those GPU buffers themselves. - When a session is created with `preferredOutputLocation` == "gpu-buffer" specified in session options, and the corresponding output is not pre-allocated, user need to call the output tensor's `dispose()` or `getData(true)` to manually release the underlying GPU buffers. - ORT internal errors (including providing a pre-allocated output tensor with wrong type/dims) will invalidate the whole wasm memory and is not recoverable. An exception is thrown in this situation.
2023-09-29 18:24:42 +00:00
return new Promise<SerializableTensorMetadata[]>((resolve, reject) => {
enqueueCallbacks('run', [resolve, reject]);
[js/webgpu] support IO binding (#17480) <del> **This PR is based on a few prerequisites PRs. They are listed as below:** - #17465 - #17469 - #17470 - #17472 - #17473 - #17484 Please review the current change by only looking at commit e2e6623e673ec6de55a5c1f8edcbd3a46b535a89 and later. </del> ### Description This PR introduces WebGPU IO binding. This new feature allows onnxruntime-web users to use tensors created from GPU as model input/output so that a model inferencing can be done without unnecessary data copy between CPU and GPU for model input/output. ### Examples An E2E demo/example is being worked on. Following is some simple demo with code snippet. Let's first check today how we do: ```js // STEP.1 - create an inference session: const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'] }); // STEP.2 - create model input: (supposing myImageCpuData is a Float32Array) const feeds = { 'input_image:0': new ort.Tensor('float32', myImageCpuData, [1, 224, 224, 3]) }; // STEP.3 - run model const myResults = await mySession.run(feeds); // STEP.4 - get output data const myData = myResults['output_image:0'].data; // Float32Array ``` #### for inputs (GPU tensor): Now, with IO binding, you can create a tensor from a GPU buffer, and feed it to the model: ```js // new STEP.2.A - create model input from a GPU buffer: (supposing myInputGpuBuffer is a `GPUBuffer` object with input data) const feeds = { 'input_image:0': ort.Tensor.fromGpuBuffer(myInputGpuBuffer, { dataType: 'float32', dims: [1, 224, 224, 3] }) }; ``` ### for outputs (pre-allocated GPU tensor) you can also do that for output, **if you know the output shape**: ```js // new STEP.2.B - create model output from a GPU buffer: (supposing myOutputGpuBuffer is a pre-allocated `GPUBuffer` object) const fetches = { 'output_image:0': ort.Tensor.fromGpuBuffer(myOutputGpuBuffer, { dataType: 'float32', dims: [1, 512, 512, 3] }) }; // new STEP.3 - run model with pre-allocated output (fetches) const myResults = await mySession.run(feeds, fetches); ``` ### for outputs (specify location) if you do not know the output shape, you can specify the output location when creating the session: ```js // new STEP.1 - create an inference session with an option "preferredOutputLocation": const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'], preferredOutputLocation: "gpu-buffer" }); ``` if the model has multiple outputs, you can specify them seperately: ```js // new STEP.1 - create an inference session with an option "preferredOutputLocation": const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'], preferredOutputLocation: { "output_image:0": "gpu-buffer" } }); ``` now you don't need to prepare the `fetches` object and onnxruntime-web will prepare output data on the location that specified. #### read data when you get the output tensor, you can: ```js // get the gpu buffer object: const gpuBuffer = myOutputTensor.gpuBuffer; // GPUBuffer // get the CPU data asynchronizely const cpuData = await myOutputTensor.getData(); // get the CPU data asynchronizely and release the underlying GPU resources const cpuData = await myOutputTensor.getData(true); // dispose the tensor (release the underlying GPU resources). This tensor object will be invalid after dispose() is called. myOutputTensor.dispose(); ``` #### resource management JavaScript has GC so you don't need to worry about managing JavaScript objects. But there are 2 types of resources that are not managed by GC: - GPU buffer that used in tensors - Underlying ORT native resources To simplify, most of the unmanaged resources and handled inside ORT web. But there are a few resources that need users to manage: - All external GPU resources, including GPU buffers inside all tensors created by `Tensor.fromGpuBuffer()`, will not be managed by ORT. User should manage those GPU buffers themselves. - When a session is created with `preferredOutputLocation` == "gpu-buffer" specified in session options, and the corresponding output is not pre-allocated, user need to call the output tensor's `dispose()` or `getData(true)` to manually release the underlying GPU buffers. - ORT internal errors (including providing a pre-allocated output tensor with wrong type/dims) will invalidate the whole wasm memory and is not recoverable. An exception is thrown in this situation.
2023-09-29 18:24:42 +00:00
const serializableInputs = inputs as SerializableTensorMetadata[]; // every input is on CPU.
const message: OrtWasmMessage =
{type: 'run', in : {sessionId, inputIndices, inputs: serializableInputs, outputIndices, options}};
proxyWorker!.postMessage(message, core.extractTransferableBuffers(serializableInputs));
});
} else {
[js/webgpu] support IO binding (#17480) <del> **This PR is based on a few prerequisites PRs. They are listed as below:** - #17465 - #17469 - #17470 - #17472 - #17473 - #17484 Please review the current change by only looking at commit e2e6623e673ec6de55a5c1f8edcbd3a46b535a89 and later. </del> ### Description This PR introduces WebGPU IO binding. This new feature allows onnxruntime-web users to use tensors created from GPU as model input/output so that a model inferencing can be done without unnecessary data copy between CPU and GPU for model input/output. ### Examples An E2E demo/example is being worked on. Following is some simple demo with code snippet. Let's first check today how we do: ```js // STEP.1 - create an inference session: const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'] }); // STEP.2 - create model input: (supposing myImageCpuData is a Float32Array) const feeds = { 'input_image:0': new ort.Tensor('float32', myImageCpuData, [1, 224, 224, 3]) }; // STEP.3 - run model const myResults = await mySession.run(feeds); // STEP.4 - get output data const myData = myResults['output_image:0'].data; // Float32Array ``` #### for inputs (GPU tensor): Now, with IO binding, you can create a tensor from a GPU buffer, and feed it to the model: ```js // new STEP.2.A - create model input from a GPU buffer: (supposing myInputGpuBuffer is a `GPUBuffer` object with input data) const feeds = { 'input_image:0': ort.Tensor.fromGpuBuffer(myInputGpuBuffer, { dataType: 'float32', dims: [1, 224, 224, 3] }) }; ``` ### for outputs (pre-allocated GPU tensor) you can also do that for output, **if you know the output shape**: ```js // new STEP.2.B - create model output from a GPU buffer: (supposing myOutputGpuBuffer is a pre-allocated `GPUBuffer` object) const fetches = { 'output_image:0': ort.Tensor.fromGpuBuffer(myOutputGpuBuffer, { dataType: 'float32', dims: [1, 512, 512, 3] }) }; // new STEP.3 - run model with pre-allocated output (fetches) const myResults = await mySession.run(feeds, fetches); ``` ### for outputs (specify location) if you do not know the output shape, you can specify the output location when creating the session: ```js // new STEP.1 - create an inference session with an option "preferredOutputLocation": const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'], preferredOutputLocation: "gpu-buffer" }); ``` if the model has multiple outputs, you can specify them seperately: ```js // new STEP.1 - create an inference session with an option "preferredOutputLocation": const mySession = await ort.InferenceSession.create('./my_model.onnx', { executionProviders: ['webgpu'], preferredOutputLocation: { "output_image:0": "gpu-buffer" } }); ``` now you don't need to prepare the `fetches` object and onnxruntime-web will prepare output data on the location that specified. #### read data when you get the output tensor, you can: ```js // get the gpu buffer object: const gpuBuffer = myOutputTensor.gpuBuffer; // GPUBuffer // get the CPU data asynchronizely const cpuData = await myOutputTensor.getData(); // get the CPU data asynchronizely and release the underlying GPU resources const cpuData = await myOutputTensor.getData(true); // dispose the tensor (release the underlying GPU resources). This tensor object will be invalid after dispose() is called. myOutputTensor.dispose(); ``` #### resource management JavaScript has GC so you don't need to worry about managing JavaScript objects. But there are 2 types of resources that are not managed by GC: - GPU buffer that used in tensors - Underlying ORT native resources To simplify, most of the unmanaged resources and handled inside ORT web. But there are a few resources that need users to manage: - All external GPU resources, including GPU buffers inside all tensors created by `Tensor.fromGpuBuffer()`, will not be managed by ORT. User should manage those GPU buffers themselves. - When a session is created with `preferredOutputLocation` == "gpu-buffer" specified in session options, and the corresponding output is not pre-allocated, user need to call the output tensor's `dispose()` or `getData(true)` to manually release the underlying GPU buffers. - ORT internal errors (including providing a pre-allocated output tensor with wrong type/dims) will invalidate the whole wasm memory and is not recoverable. An exception is thrown in this situation.
2023-09-29 18:24:42 +00:00
return core.run(sessionId, inputIndices, inputs, outputIndices, outputs, options);
}
};
export const endProfiling = async(sessionId: number): Promise<void> => {
if (!BUILD_DEFS.DISABLE_WASM_PROXY && isProxy()) {
ensureWorker();
return new Promise<void>((resolve, reject) => {
enqueueCallbacks('end-profiling', [resolve, reject]);
const message: OrtWasmMessage = {type: 'end-profiling', in : sessionId};
proxyWorker!.postMessage(message);
});
} else {
core.endProfiling(sessionId);
}
};