[js/webgpu] Enable GroupedConvVectorize path

Vectorize met 2 failed cases in a CI bot with NVIDIA GPU, but we
couldn't repro with all the GPUs at hand, including NVIDIA GPUs.
This PR introduces GPUAdapterInfo and enables this opt on non-NVIDIA
GPUs to make the bots happy.
No obivous perf gain can be seen if we enable vectorize on NVIDIA.
However, it shows big perf improvement on Intel. On my Gen12 Intel
GPU, mobilenetv2-12 perf was improved from 11.14ms to 7.1ms.

js/web/lib/wasm/jsep/backend-webgpu.ts

@@ -94,11 +94,27 @@ const getProgramInfoUniqueKey =
       return key;
     };
 
+export class AdapterInfo {
+  private vendor: string;
+
+  constructor(adapterInfo: GPUAdapterInfo) {
+    if (adapterInfo) {
+      this.vendor = adapterInfo.vendor;
+    }
+  }
+
+  // vendor could be intel, nvidia, amd, etc.
+  isVendor(vendor: string): boolean {
+    return this.vendor === vendor;
+  }
+}
+
 /**
  * this class is designed to store status and being used as a singleton for JSEP. It will be passed to jsepInit() as
  * the first parameter so that it is stored for future use.
  */
 export class WebGpuBackend {
+  adapterInfo: AdapterInfo;
   device: GPUDevice;
   /**
    * an instance of GpuDataManager to manage a GpuDataId -> GpuBuffer mapping
@@ -212,6 +228,8 @@ export class WebGpuBackend {
     }
 
     this.device = await adapter.requestDevice(deviceDescriptor);
+    const adapterInfo = await adapter.requestAdapterInfo();
+    this.adapterInfo = new AdapterInfo(adapterInfo);
     this.gpuDataManager = createGpuDataManager(this);
     this.programManager = new ProgramManager(this);
     this.kernels = new Map();

js/web/lib/wasm/jsep/init.ts

@@ -6,7 +6,7 @@ import {Env} from 'onnxruntime-common';
 import {OrtWasmModule} from '../binding/ort-wasm';
 import {DataType, getTensorElementSize} from '../wasm-common';
 
-import {WebGpuBackend} from './backend-webgpu';
+import {AdapterInfo, WebGpuBackend} from './backend-webgpu';
 import {LOG_DEBUG} from './log';
 import {TensorView} from './tensor-view';
 import {ShapeUtil} from './util';
@@ -54,6 +54,7 @@ class TensorViewImpl implements TensorView {
 }
 
 class ComputeContextImpl implements ComputeContext {
+  readonly adapterInfo: AdapterInfo;
   readonly opKernelContext: number;
   readonly inputs: readonly TensorView[];
   readonly outputCount: number;
@@ -66,6 +67,7 @@ class ComputeContextImpl implements ComputeContext {
   private customDataOffset = 0;
   private customDataSize = 0;
   constructor(private module: OrtWasmModule, private backend: WebGpuBackend, contextDataOffset: number) {
+    this.adapterInfo = backend.adapterInfo;
     const heapU32 = module.HEAPU32;
 
     // extract context data

js/web/lib/wasm/jsep/webgpu/ops/conv.ts

@@ -148,11 +148,12 @@ const conv2d = (context: ComputeContext, inputs: readonly TensorView[], attribut
   // const hasPreluActivationWeights = false; /* TODO: add support for prelu activation weights */
   const isChannelsLast = attributes.format === 'NHWC';
   if (attributes.group !== 1) {
-    // Temporarily disable createGroupedConvVectorizeProgramInfo path due to bots failures with below two cases:
+    // One CI bot with NVIDIA GPU fails with below 2 cases, but we couldn't repro them with any other GPUs, including NVIDIA ones.
     // [webgpu]Conv - conv - vectorize group - B
     // [webgpu]Conv - conv - vectorize group - D
-    const disableGroupedConvVectorize = true;
-    if (!disableGroupedConvVectorize && isChannelsLast && inputs[1].dims[0] === attributes.group &&
+    // Disable vectorize on NVIDIA to make bots happy. BTW, no obvious perf gain with vectorize is seen on NVIDIA GPUs.
+    const enableGroupedConvVectorize = context.adapterInfo.isVendor('nvidia') ? false : true;
+    if (enableGroupedConvVectorize && isChannelsLast && inputs[1].dims[0] === attributes.group &&
         inputs[1].dims[1] === 1 && attributes.dilations[0] === 1 && attributes.dilations[1] === 1) {
       const outputShape = calculateOutputShape(
           inputs[0].dims, inputs[1].dims, attributes.dilations, adjustedAttributes.pads, attributes.strides,

js/web/lib/wasm/jsep/webgpu/types.ts

@@ -2,6 +2,7 @@
 // Licensed under the MIT License.
 
 import {DataType} from '../../wasm-common';
+import {AdapterInfo} from '../backend-webgpu'
 import {TensorView} from '../tensor-view';
 
 import {ShaderHelper} from './ops/common';
@@ -146,6 +147,11 @@ export interface ComputeContextInputsOutputsMapping {
  * A ComputeContext instance carries the states that representing the current running of a kernel.
  */
 export interface ComputeContext {
+  /**
+   * gpu adapter info
+   */
+  readonly adapterInfo: AdapterInfo;
+
   /**
    * stores the pointer to OpKernelContext
    */