Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into api

Author: zhanghonggeng

.github/workflows/CheckPRTemplate.yml

@@ -31,6 +31,7 @@ jobs:
           AGILE_CHECKIN_AUTHOR: ${{ github.event.pull_request.user.login }}
           method: check_pr
           GITHUB_API_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          BRANCH: ${{ github.base_ref }}
         run: |
           python3 tools/CheckPRTemplate.py; EXCODE=$?
           echo "EXCODE: $EXCODE"

.github/workflows/_Doc-Preview.yml

@@ -55,7 +55,7 @@ jobs:
           AGILE_COMPILE_BRANCH: ${{ github.event.pull_request.base.ref }}
           BOS_CREDENTIAL_AK: "paddle"
           BOS_CREDENTIAL_SK: "paddle"
-          PADDLE_WHL: https://paddle-github-action.cdn.bcebos.com/PR/build/${{ github.event.pull_request.number }}/${{ github.event.pull_request.head.sha }}/paddlepaddle_gpu-0.0.0-cp310-cp310-linux_x86_64.whl
+          PADDLE_WHL: https://paddle-github-action.bj.bcebos.com/PR/build/${{ github.event.pull_request.number }}/${{ github.event.pull_request.head.sha }}/paddlepaddle_gpu-0.0.0-cp310-cp310-linux_x86_64.whl
         run: |
           container_name=${TASK}-$(date +%Y%m%d-%H%M%S)
           echo "container_name=${container_name}" >> ${{ github.env }}
@@ -105,6 +105,7 @@ jobs:
           curl -sS -o /tmp/entrypoint.sh https://paddle-dev-tools-open.bj.bcebos.com/fluiddoc-preview/entrypoint-paddle-docs-review.sh
           cd /
           source ${{ github.workspace }}/../../../proxy
+          pip3 install --progress-bar off -i https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple --force-reinstall ${work_dir}/build/pr_whl/*.whl
           bash "/tmp/entrypoint.sh"
           '
 

.github/workflows/remove-skip-ci-labels.yml

@@ -0,0 +1,53 @@
+name: Remove Skip-CI Labels
+
+on:
+  pull_request_target:
+    types: [synchronize]
+
+permissions:
+  pull-requests: write
+
+jobs:
+  remove-skip-ci-labels:
+    name: Remove skip-ci labels on new commits
+    runs-on: ubuntu-latest
+    steps:
+      - name: Get PR labels
+        id: get-labels
+        uses: actions/github-script@v7
+        with:
+          github-token: ${{ secrets.GITHUB_TOKEN }}
+          script: |
+            const { data: labels } = await github.rest.issues.listLabelsOnIssue({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              issue_number: context.issue.number
+            });
+
+            const skipCiLabels = labels
+              .filter(label => label.name.startsWith('skip-ci:'))
+              .map(label => label.name);
+
+            console.log('Found skip-ci labels:', skipCiLabels);
+            core.setOutput('skip-ci-labels', JSON.stringify(skipCiLabels));
+            core.setOutput('has-skip-ci-labels', skipCiLabels.length > 0 ? 'true' : 'false');
+
+      - name: Remove skip-ci labels
+        if: steps.get-labels.outputs.has-skip-ci-labels == 'true'
+        uses: actions/github-script@v7
+        with:
+          github-token: ${{ secrets.GITHUB_TOKEN }}
+          script: |
+            const skipCiLabels = JSON.parse('${{ steps.get-labels.outputs.skip-ci-labels }}');
+
+            for (const label of skipCiLabels) {
+              console.log(`Removing label: ${label}`);
+              await github.rest.issues.removeLabel({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                issue_number: context.issue.number,
+                name: label
+              });
+            }
+
+            console.log(`Successfully removed ${skipCiLabels.length} skip-ci label(s)`);

.github/workflows/rerun.yml

@@ -277,3 +277,13 @@ jobs:
           OWNER: ${{ github.repository_owner }}
           REPO: ${{ github.event.repository.name }}
           JOB_NAME: 'Api-Benchmark / Check bypass for Api-Benchmark / Check bypass'
+
+      - name: Rerun Doc-Preview
+        if: ${{ contains(github.event.comment.body, 'doc') && contains(github.event.comment.body, 'preview') }}
+        uses: ./.github/actions/rerun-workflow
+        with:
+          PR_ID: ${{ github.event.issue.number }}
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          OWNER: ${{ github.repository_owner }}
+          REPO: ${{ github.event.repository.name }}
+          JOB_NAME: 'Doc-Preview / Check bypass for Doc-Preview / Check bypass'

.github/workflows/validate-pir-versions.yml

@@ -2,14 +2,12 @@ name: Check version equality on release branch
 
 on:
   create:
-    branches:
-      - 'release/**'
   workflow_dispatch:
 
 jobs:
   check-version-and-create-issue:
     runs-on: ubuntu-latest
-    if: github.ref_type == 'branch'
+    if: github.ref_type == 'branch' && startsWith(github.ref, 'refs/heads/release/') && github.repository_owner == 'PaddlePaddle'
     steps:
       - name: Checkout code
         uses: actions/checkout@v3

cmake/configure.cmake

@@ -106,16 +106,16 @@ if(WITH_GPU)
 
   find_package(CUDA REQUIRED)
 
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_LESS 10.1)
-    message(FATAL_ERROR "Paddle needs CUDA >= 10.1 to compile")
+  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_LESS 11.7)
+    message(FATAL_ERROR "Paddle needs CUDA >= 11.7 to compile")
   endif()
 
   if(NOT CUDNN_FOUND)
     message(FATAL_ERROR "Paddle needs cudnn to compile")
   endif()
 
-  if(${CUDNN_MAJOR_VERSION} VERSION_LESS 7)
-    message(FATAL_ERROR "Paddle needs CUDNN >= 7.0 to compile")
+  if(${CUDNN_MAJOR_VERSION} VERSION_LESS 8)
+    message(FATAL_ERROR "Paddle needs CUDNN >= 8.0 to compile")
   endif()
 
   if(CUPTI_FOUND)
@@ -131,20 +131,8 @@ if(WITH_GPU)
   include_directories(${CUDA_TOOLKIT_INCLUDE})
 
   if(TENSORRT_FOUND)
-    if(WIN32)
-      if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_LESS 9)
-        message(FATAL_ERROR "TensorRT needs CUDA >= 9.0 to compile on Windows")
-      endif()
-    else()
-      if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_LESS 8)
-        message(FATAL_ERROR "TensorRT needs CUDA >= 8.0 to compile")
-      endif()
-      if(${CUDNN_MAJOR_VERSION} VERSION_LESS 7)
-        message(FATAL_ERROR "TensorRT needs CUDNN >= 7.0 to compile")
-      endif()
-      if(${TENSORRT_MAJOR_VERSION} VERSION_LESS 4)
-        message(FATAL_ERROR "Paddle needs TensorRT >= 4.0 to compile")
-      endif()
+    if(${TENSORRT_MAJOR_VERSION} VERSION_LESS 8)
+      message(FATAL_ERROR "Paddle needs TensorRT >= 8.0 to compile")
     endif()
     include_directories(${TENSORRT_INCLUDE_DIR})
   endif()

cmake/cuda.cmake

@@ -288,10 +288,6 @@ elseif(${CMAKE_CUDA_COMPILER_VERSION} LESS 13.0) # CUDA 12.0+
   set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -Wno-deprecated-gpu-targets")
 endif()
 
-if(NOT ${CMAKE_CUDA_COMPILER_VERSION} LESS 10.0)
-  add_definitions("-DTRT_PLUGIN_FP16_AVAILABLE")
-endif()
-
 add_definitions("-DCUDA_VERSION_MAJOR=\"${CUDA_VERSION_MAJOR}\"")
 add_definitions("-DCUDA_VERSION_MINOR=\"${CUDA_VERSION_MINOR}\"")
 add_definitions("-DCUDA_TOOLKIT_ROOT_DIR=\"${CUDA_TOOLKIT_ROOT_DIR}\"")

cmake/external/xpu.cmake

@@ -34,9 +34,13 @@ set(XPU_FFT_LIB_NAME "libcufft.so")
 add_compile_definitions(XPUAPI_NOT_INCLUDE_DEPRECATED)
 
 if(NOT DEFINED XPU_XHPC_BASE_DATE)
-  set(XPU_XHPC_BASE_DATE "dev/20251024")
+  set(XPU_XHPC_BASE_DATE "dev/20251103")
+endif()
+if(WITH_ARM)
+  set(XPU_XCCL_BASE_VERSION "20251104") # For XRE5
+else()
+  set(XPU_XCCL_BASE_VERSION "3.0.3.4") # For XRE5
 endif()
-set(XPU_XCCL_BASE_VERSION "3.0.3.4") # For XRE5
 if(NOT DEFINED XPU_XFT_BASE_VERSION)
   set(XPU_XFT_BASE_VERSION "20250507/xpu3")
 endif()
@@ -56,7 +60,11 @@ else()
 endif()
 
 if(NOT DEFINED XPU_FFT_BASE_DATE)
-  set(XPU_FFT_BASE_DATE "20250704")
+  if(WITH_ARM)
+    set(XPU_FFT_BASE_DATE "20251017")
+  else()
+    set(XPU_FFT_BASE_DATE "20250704")
+  endif()
 endif()
 
 set(XPU_XRE_BASE_URL
@@ -97,7 +105,7 @@ if(WITH_ARM)
   if(WITH_XPU_XRE5)
     set(XPU_XRE_DIR_NAME "xre-kylin_v10_server-aarch64-${XPU_XRE_BASE_VERSION}")
     # TODO: xccl has no kylin output now. set default value here.
-    set(XPU_XCCL_DIR_NAME "xccl_Linux_x86_64")
+    set(XPU_XCCL_DIR_NAME "xccl_Linux_aarch64")
     set(XPU_XHPC_DIR_NAME "xhpc-kylinv4_aarch64")
     set(XPU_XFT_DIR_NAME "") # TODO: xft has no kylin output at now.
   else()
@@ -106,7 +114,8 @@ if(WITH_ARM)
     set(XPU_XFT_DIR_NAME "") # TODO: xft has no kylin output at now.
   endif()
   if(WITH_XPU_FFT)
-    set(XPU_FFT_DIR_NAME "kylin_v10_aarch64/xpufft_kylinv10_aarch64_v2")
+    set(XPU_FFT_DIR_NAME
+        "kylin_v10_aarch64/${XPU_FFT_BASE_DATE}/xpufft_kylinv10_aarch64")
   endif()
 elseif(WITH_SUNWAY)
   set(XPU_XRE_DIR_NAME "xre-deepin_sw6_64")

cmake/operators.cmake

@@ -122,7 +122,7 @@ function(register_onednn_kernel TARGET)
   if(${onednn_cc_srcs_len} EQUAL 0)
     message(
       FATAL_ERROR
-        "The MKLDNN kernel file of ${TARGET} should contains at least one *.*_onednn_op.cc file"
+        "The OneDNN kernel file of ${TARGET} should contains at least one *.*_onednn_op.cc file"
     )
   endif()
   if(WITH_ONEDNN)
@@ -138,7 +138,7 @@ function(register_onednn_kernel TARGET)
     set(op_name "")
     find_register(${onednn_src} "REGISTER_OP_KERNEL" op_name)
     if(NOT ${op_name} EQUAL "")
-      file(APPEND ${pybind_file} "USE_OP_DEVICE_KERNEL(${op_name}, MKLDNN);\n")
+      file(APPEND ${pybind_file} "USE_OP_DEVICE_KERNEL(${op_name}, ONEDNN);\n")
     endif()
   endforeach()
 endfunction()
@@ -607,9 +607,9 @@ function(op_library TARGET)
     endforeach()
   endif()
 
-  # pybind USE_OP_DEVICE_KERNEL for MKLDNN
+  # pybind USE_OP_DEVICE_KERNEL for ONEDNN
   if(WITH_ONEDNN AND ${onednn_cc_srcs_len} GREATER 0)
-    # Append first implemented MKLDNN activation operator
+    # Append first implemented ONEDNN activation operator
     if(${ONEDNN_FILE} STREQUAL "activation_onednn_op")
       file(APPEND ${pybind_file} "USE_OP_DEVICE_KERNEL(softplus, MKLDNN);\n")
     else()

paddle/ap/include/axpr/anf_expr_util.h

@@ -458,7 +458,7 @@ struct ParseJsonToAnfExprHelperInt64 {
     if (!j_obj.is_number_integer()) {
       return JsonParseMismatch(j_obj,
                                "ParseJsonToAnfExpr<int64_t>: json object "
-                               "should be a intergral number.");
+                               "should be a integral number.");
     }
     auto c = j_obj.get<Json::number_integer_t>();
     return AnfExpr{AnfExprBuilder().Int64(c)};