Merge branch 'main' into yifanl/trt_ver_update_1

.github/workflows/cffconvert.yml

@@ -8,7 +8,7 @@ on:
 jobs:
   validate:
     name: "validate"
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - name: Check out a copy of the repository
         uses: actions/checkout@v4

.github/workflows/codeql.yml

@@ -18,7 +18,7 @@ on:
 jobs:
   analyze:
     name: Analyze
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     permissions:
       actions: read
       contents: read
@@ -55,6 +55,11 @@ jobs:
         java-version: '11'
         distribution: 'microsoft'
 
+    - if: ${{ matrix.language == 'javascript' }}
+      uses: actions/setup-node@v4
+      with:
+        node-version: 20
+
     # Autobuild attempts to build any compiled languages  (C/C++, C#, or Java).
     # If this step fails, then you should remove it and run the build manually (see below)
     - if: ${{ matrix.language != 'cpp' }}

.github/workflows/gradle-wrapper-validation.yml

@@ -8,7 +8,7 @@ on: [push, pull_request]
 jobs:
   validation:
     name: "Validation"
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - uses: actions/checkout@v4
       - uses: gradle/actions/wrapper-validation@v4

.github/workflows/labeler.yml

@@ -8,7 +8,7 @@ permissions:
 
 jobs:
   triage:
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
     - uses: github/[email protected]
       with:

.github/workflows/lint.yml

@@ -36,7 +36,7 @@ jobs:
   lint-python-format:
     # Required workflow
     name: Python format
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - uses: actions/checkout@v4
       - name: Setup Python
@@ -114,9 +114,12 @@ jobs:
 
   lint-js:
     name: Lint JavaScript
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - uses: actions/checkout@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 20
       - uses: reviewdog/action-eslint@v1
         with:
           reporter: github-pr-check

.github/workflows/linux_training.yml

@@ -12,15 +12,15 @@ concurrency:
 
 jobs:
   orttraining-linux-ci-pipeline:
-    runs-on: ubuntu-24.04
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     permissions:
       actions: read
       contents: read
       security-events: write
     steps:
       - uses: actions/checkout@v4
       - run: |
-         python3 -m pip install -r tools/ci_build/github/linux/python/requirements.txt
+         python3 -m pip install --user -r tools/ci_build/github/linux/python/requirements.txt
       - name: Initialize CodeQL
         uses: github/codeql-action/init@v3
         with:

.github/workflows/pr_checks.yml

@@ -19,7 +19,7 @@ concurrency:
 jobs:
   auto-apply-fixes:
     name: Suggest fixes
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     permissions:
       contents: read
       pull-requests: write
@@ -34,14 +34,13 @@ jobs:
         with:
           toolchain: stable
           components: rustfmt
-      - name: Install dependencies
+      - name: Install dependencies and run lintrunner on all files
         run: |
-          python -m pip install -r requirements-dev.txt
-          python -m pip install lintrunner lintrunner-adapters
+	  export PATH=$HOME/.local/bin:$PATH
+	  set -e
+          python -m pip install --user -r requirements-dev.txt
+          python -m pip install --user lintrunner lintrunner-adapters
           lintrunner init
-      - name: Run lintrunner on all files
-        run: |
-          set +e
           lintrunner f --all-files -v
           exit 0
       - uses: parkerbxyz/suggest-changes@v1

.github/workflows/publish-c-apidocs.yml

@@ -22,7 +22,7 @@ permissions:
 jobs:
   build:
     name: Generate C/C++ API docs
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - uses: actions/checkout@v4
       - name: Install doxygen and dependencies

.github/workflows/publish-csharp-apidocs.yml

@@ -20,7 +20,7 @@ permissions:
 
 jobs:
   build:
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     env:
       DOCFXVERSION: 2.62.2
     steps:

.github/workflows/publish-gh-pages.yml

@@ -8,7 +8,7 @@ on:
 
 jobs:
   placeholder:
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - name: Placeholder step to have workflow included in the GitHub web UI
         run: |

.github/workflows/publish-java-apidocs.yml

@@ -21,7 +21,7 @@ permissions:
 jobs:
   build:
     name: Generate Java docs
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - uses: actions/checkout@v4
       - name: Set up JDK 11

.github/workflows/publish-js-apidocs.yml

@@ -21,7 +21,7 @@ permissions:
 jobs:
   build:
     name: Generate JS API docs
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - uses: actions/checkout@v4
       - name: Setup Node.js

.github/workflows/publish-python-apidocs.yml

@@ -22,7 +22,7 @@ permissions:
 jobs:
   build:
     name: Generate Python API docs
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - uses: actions/checkout@v4
       - name: Install tools

.github/workflows/skip-doc-change.yml.j2

@@ -14,7 +14,7 @@ jobs:
 {%- for name in job_names %}
   job{{ loop.index }}:
     name: {{ name }}
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
       - run: 'echo "No build required, only documentation changed"'
 {% endfor %}

.github/workflows/stale.yml

@@ -8,7 +8,7 @@ on:
 
 jobs:
   close-stale-issues:
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     permissions:
       issues: write
       pull-requests: write

.github/workflows/title-only-labeler.yml

@@ -8,7 +8,7 @@ permissions:
 
 jobs:
   triage:
-    runs-on: ubuntu-latest
+    runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-Ubuntu2204-AMD-CPU"]
     steps:
     - uses: github/[email protected]
       with:

js/node/README.md

@@ -14,7 +14,7 @@ Refer to [ONNX Runtime JavaScript examples](https://github.com/microsoft/onnxrun
 
 ## Requirements
 
-ONNXRuntime works on Node.js v16.x+ (recommend v18.x+) or Electron v15.x+ (recommend v28.x+).
+ONNXRuntime works on Node.js v16.x+ (recommend v20.x+) or Electron v15.x+ (recommend v28.x+).
 
 The following table lists the supported versions of ONNX Runtime Node.js binding provided with pre-built binaries.
 

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

@@ -902,6 +902,10 @@ export class WebGpuBackend {
     this.sessionStatus = 'default';
   }
 
+  onCreateSession(): void {
+    this.gpuDataManager.onCreateSession();
+  }
+
   onReleaseSession(sessionId: number): void {
     this.unregisterBuffers(sessionId);
     if (this.capturedCommandList.has(sessionId)) {

js/web/lib/wasm/jsep/webgpu/gpu-data-manager.ts

@@ -64,6 +64,11 @@ export interface GpuDataManager {
    */
   dispose(): void;
 
+  /**
+   * create session related data.
+   */
+  onCreateSession(): void;
+
   /**
    * release session related data.
    * @param sessionId - specify the session ID.
@@ -200,6 +205,9 @@ class GpuDataManagerImpl implements GpuDataManager {
   // a SessionID -> GPUBuffer[] mapping.
   private capturedPendingBuffers: Map<number, GPUBuffer[]>;
 
+  // The session count.
+  private sessionCount: number;
+
   constructor(private backend: WebGpuBackend) {
     this.storageCache = new Map();
     this.freeBuffers = new Map();
@@ -213,6 +221,8 @@ class GpuDataManagerImpl implements GpuDataManager {
       this.freeBuffers.set(key, []);
       this.freeUniformBuffers.set(key, []);
     }
+
+    this.sessionCount = 0;
   }
 
   upload(id: GpuDataId, data: Uint8Array): void {
@@ -360,7 +370,12 @@ class GpuDataManagerImpl implements GpuDataManager {
   release(id: GpuDataId): number {
     const cachedData = this.storageCache.get(id);
     if (!cachedData) {
-      throw new Error('releasing data does not exist');
+      if (this.storageCache.size === 0) {
+        // cache was previously cleared, no need to release anything.
+        return 0;
+      } else {
+        throw new Error('releasing data does not exist');
+      }
     }
 
     LOG_DEBUG('verbose', () => `[WebGPU] GpuDataManager.release(id=${id}), gpuDataId=${cachedData.gpuData.id}`);
@@ -460,6 +475,10 @@ class GpuDataManagerImpl implements GpuDataManager {
     this.capturedPendingBuffers = new Map();
   }
 
+  onCreateSession() {
+    this.sessionCount += 1;
+  }
+
   onReleaseSession(sessionId: number) {
     // release the captured pending buffers.
     const pendingBuffers = this.capturedPendingBuffers.get(sessionId);
@@ -469,6 +488,16 @@ class GpuDataManagerImpl implements GpuDataManager {
       });
       this.capturedPendingBuffers.delete(sessionId);
     }
+
+    // release the storage cache if no active sessions.
+    this.sessionCount -= 1;
+    if (this.sessionCount === 0) {
+      LOG_DEBUG('warning', () => '[WebGPU] Clearing webgpu buffer cache');
+      this.storageCache.forEach((storage) => {
+        storage.gpuData.buffer.destroy();
+      });
+      this.storageCache = new Map();
+    }
   }
 }
 

js/web/lib/wasm/wasm-core-impl.ts

@@ -317,6 +317,8 @@ export const createSession = async (
       checkLastError("Can't create a session.");
     }
 
+    wasm.jsepOnCreateSession?.();
+
     // clear current MLContext after session creation
     if (wasm.currentContext) {
       wasm.jsepRegisterMLContext!(sessionHandle, wasm.currentContext);

js/web/lib/wasm/wasm-types.ts

@@ -141,6 +141,12 @@ export declare namespace JSEP {
      * @param sessionId - specify the session ID.
      */
     jsepOnRunStart: (sessionId: number) => void;
+    /**
+     * [exported from pre-jsep.js] Create a session. This function will be called after _OrtCreateSession() is
+     * called.
+     * @returns
+     */
+    jsepOnCreateSession: () => void;
     /**
      * [exported from pre-jsep.js] Release a session. This function will be called before _OrtReleaseSession() is
      * called.

onnxruntime/core/providers/webnn/builders/impl/expand_op_builder.cc

@@ -95,11 +95,6 @@ bool ExpandOpBuilder::IsOpSupportedImpl(const InitializedTensorSet& initializers
     return false;
   }
 
-  if (input_shape.empty()) {
-    LOGS(logger, VERBOSE) << "Expand does not support empty input's shape.";
-    return false;
-  }
-
   std::vector<int64_t> output_shape;
   if (!GetBidirectionalBroadcastShape(input_shape, new_shape, output_shape)) {
     LOGS(logger, VERBOSE) << "The input cannot expand to shape " << GetShapeString(new_shape);