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

Author: AndSonder

.github/workflows/Build-develop.yml

@@ -53,11 +53,13 @@ jobs:
           CUDA_VERSION:
           CUDNN_VERSION:
           WITH_XPU_BKCL: "ON"
+          WITH_XPU_FFT: "ON"
           WITH_XPU_XRE5: "ON"
           CACHE_DIR: /root/.cache
           CCACHE_DIR: /root/.ccache
           CCACHE_MAXSIZE: 150G
           CCACHE_LIMIT_MULTIPLE: 0.8
+          IF_KUNLUN3: "ON"
           GITHUB_API_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           home_dir: ${{ github.workspace }}/../../../..
         run: |
@@ -85,6 +87,7 @@ jobs:
             -e CUDA_VERSION \
             -e CUDNN_VERSION \
             -e WITH_XPU_BKCL \
+            -e WITH_XPU_FFT \
             -e WITH_XPU_XRE5 \
             -e WITH_INFERENCE_API_TEST \
             -e CACHE_DIR \
@@ -93,6 +96,7 @@ jobs:
             -e CCACHE_LIMIT_MULTIPLE \
             -e ci_scripts \
             -e WITH_AVX \
+            -e IF_KUNLUN3 \
             -e no_proxy \
             -e GITHUB_API_TOKEN \
             -e CI_name \
@@ -212,6 +216,7 @@ jobs:
           CCACHE_DIR: /root/.ccache
           CCACHE_MAXSIZE: 150G
           CCACHE_LIMIT_MULTIPLE: 0.8
+          IF_KUNLUN3: "ON"
           GITHUB_API_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           home_dir: ${{ github.workspace }}/../../../..
           FLAGS_use_stride_kernel: "0"
@@ -252,6 +257,7 @@ jobs:
             -e CCACHE_LIMIT_MULTIPLE \
             -e ci_scripts \
             -e WITH_AVX \
+            -e IF_KUNLUN3 \
             -e no_proxy \
             -e GITHUB_API_TOKEN \
             -e FLAGS_use_stride_kernel \

.github/workflows/_Linux-XPU.yml

@@ -177,15 +177,12 @@ set +x
             echo "Training Resnet50 completed!"
 
             #inference Reset50
-            IFS=',' read -ra DEVICES <<< "$CUDA_VISIBLE_DEVICES"
-            echo ${DEVICES[0]}
-
             echo "Starting to predict ResNet50 model..."
             python main.py -c paddlex/configs/modules/image_classification/ResNet50.yaml \
                 -o Global.mode=predict \
                 -o Predict.model_dir="./resnet50_output/best_model/inference" \
                 -o Predict.input="https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_image_classification_001.jpg" \
-                -o Global.device="xpu:${DEVICES[0]}"
+                -o Global.device="xpu:0"
             echo "Predicting Resnet50 completed!"
             cd ..
             export FLAGS_enable_pir_api=1

ci/kunlun_test.sh

@@ -216,13 +216,13 @@ function run_setup(){
     echo "if you use setup.py to compile,please export envs as following in /paddle ..."
     cat << EOF
     ========================================
-    export CMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release} WITH_GPU=${WITH_GPU:-OFF} WITH_SHARED_PHI=${WITH_SHARED_PHI:-OFF} WITH_TENSORRT=${WITH_TENSORRT:-ON} WITH_OPENVINO=${WITH_OPENVINO:-OFF} WITH_ROCM=${WITH_ROCM:-OFF} WITH_CINN=${WITH_CINN:-OFF} WITH_DISTRIBUTE=${distributed_flag} WITH_MKL=${WITH_MKL:-ON} WITH_AVX=${WITH_AVX:-OFF} CUDA_ARCH_NAME=${CUDA_ARCH_NAME:-All} NEW_RELEASE_PYPI=${NEW_RELEASE_PYPI:-OFF} NEW_RELEASE_ALL=${NEW_RELEASE_ALL:-OFF} NEW_RELEASE_JIT=${NEW_RELEASE_JIT:-OFF} WITH_PYTHON=${WITH_PYTHON:-ON} CUDNN_ROOT=/usr/ WITH_TESTING=${WITH_TESTING:-ON} WITH_COVERAGE=${WITH_COVERAGE:-OFF} WITH_INCREMENTAL_COVERAGE=${WITH_INCREMENTAL_COVERAGE:-OFF} CMAKE_MODULE_PATH=/opt/rocm/hip/cmake CMAKE_EXPORT_COMPILE_COMMANDS=ON WITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} INFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} PY_VERSION=${PY_VERSION:-3.8} CMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} WITH_PSCORE=${pscore_flag} WITH_PSLIB=${pslib_flag} WITH_GLOO=${gloo_flag} WITH_XPU=${WITH_XPU:-OFF} WITH_IPU=${WITH_IPU:-OFF} XPU_SDK_ROOT=${XPU_SDK_ROOT:-""} WITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} WITH_XPU_XRE5=${WITH_XPU_XRE5:-OFF} WITH_ARM=${WITH_ARM:-OFF} WITH_STRIP=${WITH_STRIP:-ON} ON_INFER=${ON_INFER:-OFF} WITH_HETERPS=${WITH_HETERPS:-OFF} CUDA_ARCH_BIN=${CUDA_ARCH_BIN} WITH_RECORD_BUILDTIME=${WITH_RECORD_BUILDTIME:-OFF} WITH_UNITY_BUILD=${WITH_UNITY_BUILD:-OFF} WITH_ONNXRUNTIME=${WITH_ONNXRUNTIME:-OFF} WITH_CUDNN_FRONTEND=${WITH_CUDNN_FRONTEND:-OFF} WITH_CPP_TEST=${WITH_CPP_TEST:-OFF} FA_BUILD_WITH_CACHE=${FA_BUILD_WITH_CACHE:-ON}
+    export CMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release} WITH_GPU=${WITH_GPU:-OFF} WITH_SHARED_PHI=${WITH_SHARED_PHI:-OFF} WITH_TENSORRT=${WITH_TENSORRT:-ON} WITH_OPENVINO=${WITH_OPENVINO:-OFF} WITH_ROCM=${WITH_ROCM:-OFF} WITH_CINN=${WITH_CINN:-OFF} WITH_DISTRIBUTE=${distributed_flag} WITH_MKL=${WITH_MKL:-ON} WITH_AVX=${WITH_AVX:-OFF} CUDA_ARCH_NAME=${CUDA_ARCH_NAME:-All} NEW_RELEASE_PYPI=${NEW_RELEASE_PYPI:-OFF} NEW_RELEASE_ALL=${NEW_RELEASE_ALL:-OFF} NEW_RELEASE_JIT=${NEW_RELEASE_JIT:-OFF} WITH_PYTHON=${WITH_PYTHON:-ON} CUDNN_ROOT=/usr/ WITH_TESTING=${WITH_TESTING:-ON} WITH_COVERAGE=${WITH_COVERAGE:-OFF} WITH_INCREMENTAL_COVERAGE=${WITH_INCREMENTAL_COVERAGE:-OFF} CMAKE_MODULE_PATH=/opt/rocm/hip/cmake CMAKE_EXPORT_COMPILE_COMMANDS=ON WITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} INFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} PY_VERSION=${PY_VERSION:-3.8} CMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} WITH_PSCORE=${pscore_flag} WITH_PSLIB=${pslib_flag} WITH_GLOO=${gloo_flag} WITH_XPU=${WITH_XPU:-OFF} WITH_IPU=${WITH_IPU:-OFF} XPU_SDK_ROOT=${XPU_SDK_ROOT:-""} WITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} WITH_XPU_XHPC=${WITH_XPU_XHPC:-OFF} WITH_XPU_XFT=${WITH_XPU_XFT:-OFF} WITH_XPU_XRE5=${WITH_XPU_XRE5:-OFF} WITH_XPU_FFT=${WITH_XPU_FFT:-OFF} WITH_ARM=${WITH_ARM:-OFF} WITH_STRIP=${WITH_STRIP:-ON} ON_INFER=${ON_INFER:-OFF} WITH_HETERPS=${WITH_HETERPS:-OFF} CUDA_ARCH_BIN=${CUDA_ARCH_BIN} WITH_RECORD_BUILDTIME=${WITH_RECORD_BUILDTIME:-OFF} WITH_UNITY_BUILD=${WITH_UNITY_BUILD:-OFF} WITH_ONNXRUNTIME=${WITH_ONNXRUNTIME:-OFF} WITH_CUDNN_FRONTEND=${WITH_CUDNN_FRONTEND:-OFF} WITH_CPP_TEST=${WITH_CPP_TEST:-OFF} FA_BUILD_WITH_CACHE=${FA_BUILD_WITH_CACHE:-ON}
     ========================================
 EOF
     echo "if you use cmake to compile,please Configuring cmake in /paddle/build ..."
     cat <<EOF
     ========================================
-    cmake .. -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release} -DWITH_GPU=${WITH_GPU:-OFF} -DWITH_SHARED_PHI=${WITH_SHARED_PHI:-OFF} -DWITH_TENSORRT=${WITH_TENSORRT:-ON} -DWITH_OPENVINO=${WITH_OPENVINO:-OFF} -DWITH_ROCM=${WITH_ROCM:-OFF} -DWITH_CINN=${WITH_CINN:-OFF} -DWITH_DISTRIBUTE=${distributed_flag} -DWITH_MKL=${WITH_MKL:-ON} -DWITH_AVX=${WITH_AVX:-OFF} -DCUDA_ARCH_NAME=${CUDA_ARCH_NAME:-All} -DNEW_RELEASE_PYPI=${NEW_RELEASE_PYPI:-OFF} -DNEW_RELEASE_ALL=${NEW_RELEASE_ALL:-OFF} -DNEW_RELEASE_JIT=${NEW_RELEASE_JIT:-OFF} -DWITH_PYTHON=${WITH_PYTHON:-ON} -DCUDNN_ROOT=/usr/ -DWITH_TESTING=${WITH_TESTING:-ON} -DWITH_COVERAGE=${WITH_COVERAGE:-OFF} -DWITH_INCREMENTAL_COVERAGE=${WITH_INCREMENTAL_COVERAGE:-OFF} -DCMAKE_MODULE_PATH=/opt/rocm/hip/cmake -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DWITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} -DINFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} -DPY_VERSION=${PY_VERSION:-3.8} -DCMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} -DWITH_PSCORE=${pscore_flag} -DWITH_PSLIB=${pslib_flag} -DWITH_GLOO=${gloo_flag} -DWITH_XPU=${WITH_XPU:-OFF} -DWITH_IPU=${WITH_IPU:-OFF} -DXPU_SDK_ROOT=${XPU_SDK_ROOT:-""} -DWITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} -DWITH_XPU_XRE5=${WITH_XPU_XRE5:-OFF} -DWITH_ARM=${WITH_ARM:-OFF} -DWITH_STRIP=${WITH_STRIP:-ON} -DON_INFER=${ON_INFER:-OFF} -DWITH_HETERPS=${WITH_HETERPS:-OFF} -DCUDA_ARCH_BIN=${CUDA_ARCH_BIN} -DWITH_RECORD_BUILDTIME=${WITH_RECORD_BUILDTIME:-OFF} -DWITH_UNITY_BUILD=${WITH_UNITY_BUILD:-OFF} -DWITH_ONNXRUNTIME=${WITH_ONNXRUNTIME:-OFF} -DWITH_CUDNN_FRONTEND=${WITH_CUDNN_FRONTEND:-OFF} -DWITH_CPP_TEST=${WITH_CPP_TEST:-OFF} -DFA_BUILD_WITH_CACHE=${FA_BUILD_WITH_CACHE:-ON}
+    cmake .. -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release} -DWITH_GPU=${WITH_GPU:-OFF} -DWITH_SHARED_PHI=${WITH_SHARED_PHI:-OFF} -DWITH_TENSORRT=${WITH_TENSORRT:-ON} -DWITH_OPENVINO=${WITH_OPENVINO:-OFF} -DWITH_ROCM=${WITH_ROCM:-OFF} -DWITH_CINN=${WITH_CINN:-OFF} -DWITH_DISTRIBUTE=${distributed_flag} -DWITH_MKL=${WITH_MKL:-ON} -DWITH_AVX=${WITH_AVX:-OFF} -DCUDA_ARCH_NAME=${CUDA_ARCH_NAME:-All} -DNEW_RELEASE_PYPI=${NEW_RELEASE_PYPI:-OFF} -DNEW_RELEASE_ALL=${NEW_RELEASE_ALL:-OFF} -DNEW_RELEASE_JIT=${NEW_RELEASE_JIT:-OFF} -DWITH_PYTHON=${WITH_PYTHON:-ON} -DCUDNN_ROOT=/usr/ -DWITH_TESTING=${WITH_TESTING:-ON} -DWITH_COVERAGE=${WITH_COVERAGE:-OFF} -DWITH_INCREMENTAL_COVERAGE=${WITH_INCREMENTAL_COVERAGE:-OFF} -DCMAKE_MODULE_PATH=/opt/rocm/hip/cmake -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DWITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} -DINFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} -DPY_VERSION=${PY_VERSION:-3.8} -DCMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} -DWITH_PSCORE=${pscore_flag} -DWITH_PSLIB=${pslib_flag} -DWITH_GLOO=${gloo_flag} -DWITH_XPU=${WITH_XPU:-OFF} -DWITH_IPU=${WITH_IPU:-OFF} -DXPU_SDK_ROOT=${XPU_SDK_ROOT:-""} -DWITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} -DWITH_XPU_XHPC=${WITH_XPU_XHPC:-OFF} -DWITH_XPU_XFT=${WITH_XPU_XFT:-OFF} -DWITH_XPU_XRE5=${WITH_XPU_XRE5:-OFF} -DWITH_XPU_FFT=${WITH_XPU_FFT:-OFF} -DWITH_ARM=${WITH_ARM:-OFF} -DWITH_STRIP=${WITH_STRIP:-ON} -DON_INFER=${ON_INFER:-OFF} -DWITH_HETERPS=${WITH_HETERPS:-OFF} -DCUDA_ARCH_BIN=${CUDA_ARCH_BIN} -DWITH_RECORD_BUILDTIME=${WITH_RECORD_BUILDTIME:-OFF} -DWITH_UNITY_BUILD=${WITH_UNITY_BUILD:-OFF} -DWITH_ONNXRUNTIME=${WITH_ONNXRUNTIME:-OFF} -DWITH_CUDNN_FRONTEND=${WITH_CUDNN_FRONTEND:-OFF} -DWITH_CPP_TEST=${WITH_CPP_TEST:-OFF} -DFA_BUILD_WITH_CACHE=${FA_BUILD_WITH_CACHE:-ON}
     ========================================
 EOF
     export CMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release}
@@ -256,7 +256,10 @@ EOF
     export WITH_IPU=${WITH_IPU:-OFF}
     export XPU_SDK_ROOT=${XPU_SDK_ROOT:-""}
     export WITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF}
+    export WITH_XPU_XHPC=${WITH_XPU_XHPC:-OFF}
+    export WITH_XPU_XFT=${WITH_XPU_XFT:-OFF}
     export WITH_XPU_XRE5=${WITH_XPU_XRE5:-OFF}
+    export WITH_XPU_FFT=${WITH_XPU_FFT:-OFF}
     export WITH_ARM=${WITH_ARM:-OFF}
     export WITH_STRIP=${WITH_STRIP:-ON}
     export ON_INFER=${ON_INFER:-OFF}

ci/run_setup.sh

@@ -479,7 +479,10 @@ function cmake_base() {
         -DWITH_IPU=${WITH_IPU:-OFF}
         -DWITH_UNITY_BUILD=${WITH_UNITY_BUILD:-OFF}
         -DWITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF}
+        -DWITH_XPU_XHPC=${WITH_XPU_XHPC:-OFF}
+        -DWITH_XPU_XFT=${WITH_XPU_XFT:-OFF}
         -DWITH_XPU_XRE5=${WITH_XPU_XRE5:-OFF}
+        -DWITH_XPU_FFT=${WITH_XPU_FFT:-OFF}
         -DWITH_ARM=${WITH_ARM:-OFF}
         -DWITH_STRIP=${WITH_STRIP:-ON}
         -DON_INFER=${ON_INFER:-OFF}
@@ -531,6 +534,9 @@ EOF
         -DWITH_IPU=${WITH_IPU:-OFF} \
         -DXPU_SDK_ROOT=${XPU_SDK_ROOT:-""} \
         -DWITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} \
+        -DWITH_XPU_XHPC=${WITH_XPU_XHPC:-OFF} \
+        -DWITH_XPU_XFT=${WITH_XPU_XFT:-OFF} \
+        -DWITH_XPU_FFT=${WITH_XPU_FFT:-OFF} \
         -DWITH_ARM=${WITH_ARM:-OFF} \
         -DWITH_STRIP=${WITH_STRIP:-ON} \
         -DON_INFER=${ON_INFER:-OFF} \

ci/utils.sh

@@ -225,6 +225,14 @@ std::vector<paddle::Tensor> RunBackward(
   std::unordered_map<GradNodeBase*, int> node_in_degree_map =
       getInDegreeMap(queue);
 
+  std::deque<GradNodeBase*> ready_queue;
+  for (GradNodeBase* item : queue) {
+    if (!node_in_degree_map.count(item)) {
+      ready_queue.push_back(item);
+    }
+  }
+  queue = ready_queue;
+
   std::list<GradNodeBase*> force_sequential_nodes_forward_queue =
       egr::Controller::Instance().GetForceSequentialNodes();
   std::deque<GradNodeBase*> force_sequential_nodes_queue;
@@ -256,10 +264,6 @@ std::vector<paddle::Tensor> RunBackward(
     GradNodeBase* node = queue.front();
     VLOG(3) << "Preparing GradNode:" << node->name() << " addr:" << node;
     try {
-      if (queue.size() > 1 && node_in_degree_map[node] != 0) {
-        queue.pop_front();
-        continue;
-      }
       queue.pop_front();
 
       // Run node: This is where Hook happens

paddle/fluid/eager/backward.cc

@@ -141,6 +141,8 @@
     'Pool2DInferMeta',
     'ReduceIntArrayAxisInferMetaBase',
     'ReduceIntArrayAxisInferMeta',
+    'StrictReduceIntArrayAxisInferMetaBase',
+    'StrictReduceIntArrayAxisInferMeta',
     'ReshapeInferMeta',
     'ReshapeWithXShapeInferMeta',
     'ReverseInferMeta',

paddle/fluid/pir/dialect/op_generator/op_build_gen.py

@@ -1278,10 +1278,6 @@ paddle::optional<paddle::Tensor> GetOptionalTensorFromArgs(
     const phi::distributed::ProcessMesh* mesh) {
   PyObject* obj = PyTuple_GET_ITEM(args, arg_idx);
 
-  if (PyTuple_Check(obj)) {
-    obj = PyTuple_GET_ITEM(obj, 0);
-  }
-
   if (obj == nullptr || obj == Py_None) {
     if (!dispensable) {
       PADDLE_THROW(common::errors::InvalidArgument(
@@ -1611,10 +1607,6 @@ paddle::Tensor* GetTensorPtrFromArgs(const std::string& op_type,
                                      bool dispensable) {
   PyObject* obj = PyTuple_GET_ITEM(args, arg_idx);
 
-  if (PyTuple_Check(obj)) {
-    obj = PyTuple_GET_ITEM(obj, 0);
-  }
-
   if (obj == nullptr || obj == Py_None) {
     if (!dispensable) {
       PADDLE_THROW(common::errors::InvalidArgument(