[benchmark] add cmake graph (#51222)

* [benchmark] add cmake graph

* [benchmark] add cmake graph

* [benchmark] set gpu_graph off defaultly

paddle/scripts/paddle_build.sh

@@ -245,6 +245,7 @@ function cmake_base() {
         -DWITH_STRIP=${WITH_STRIP:-ON}
         -DON_INFER=${ON_INFER:-OFF}
         -DWITH_HETERPS=${WITH_HETERPS:-OFF}
+        -DWITH_GPU_GRAPH=${WITH_GPU_GRAPH:-OFF}
         -DWITH_FLUID_ONLY=${WITH_FLUID_ONLY:-OFF}
         -DWITH_RECORD_BUILDTIME=${WITH_RECORD_BUILDTIME:-OFF}
         -DCUDA_ARCH_BIN="${CUDA_ARCH_BIN}"
@@ -302,6 +303,7 @@ EOF
         -DWITH_STRIP=${WITH_STRIP:-ON} \
         -DON_INFER=${ON_INFER:-OFF} \
         -DWITH_HETERPS=${WITH_HETERPS:-OFF} \
+        -DWITH_GPU_GRAPH=${WITH_GPU_GRAPH:-OFF} \
         -DWITH_FLUID_ONLY=${WITH_FLUID_ONLY:-OFF} \
         -DCUDA_ARCH_BIN="${CUDA_ARCH_BIN}" \
         -DWITH_RECORD_BUILDTIME=${WITH_RECORD_BUILDTIME:-OFF} \
@@ -3638,13 +3640,13 @@ 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_TENSORRT=${WITH_TENSORRT:-ON} WITH_ROCM=${WITH_ROCM:-OFF} WITH_CINN=${WITH_CINN:-OFF} WITH_DISTRIBUTE=${distibuted_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_CONTRIB=${WITH_CONTRIB:-ON} WITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} WITH_INFRT=${WITH_INFRT:-OFF} INFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} PY_VERSION=${PY_VERSION:-3.7} CMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} WITH_PSCORE=${pscore_flag} WITH_PSLIB=${pslib_flag} WITH_GLOO=${gloo_flag} LITE_GIT_TAG=release/v2.10 WITH_XPU=${WITH_XPU:-OFF} WITH_MLU=${WITH_MLU:-OFF} WITH_IPU=${WITH_IPU:-OFF} WITH_CNCL=${WITH_CNCL:-OFF} XPU_SDK_ROOT=${XPU_SDK_ROOT:-""} WITH_LITE=${WITH_LITE:-OFF} WITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} WITH_ARM=${WITH_ARM:-OFF} WITH_ASCEND=${WITH_ASCEND:-OFF} WITH_ASCEND_CL=${WITH_ASCEND_CL:-OFF} WITH_ASCEND_INT64=${WITH_ASCEND_INT64:-OFF} WITH_STRIP=${WITH_STRIP:-ON} ON_INFER=${ON_INFER:-OFF} WITH_HETERPS=${WITH_HETERPS:-OFF} WITH_FLUID_ONLY=${WITH_FLUID_ONLY:-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=${WITH_CUDNN_FRONTEND:-OFF}
+    export CMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release} WITH_GPU=${WITH_GPU:-OFF} WITH_TENSORRT=${WITH_TENSORRT:-ON} WITH_ROCM=${WITH_ROCM:-OFF} WITH_CINN=${WITH_CINN:-OFF} WITH_DISTRIBUTE=${distibuted_flag} WITH_MKL=${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=${WITH_INCREMENTAL_COVERAGE:-OFF} CMAKE_MODULE_PATH=/opt/rocm/hip/cmake CMAKE_EXPORT_COMPILE_COMMANDS=ON WITH_CONTRIB=${WITH_CONTRIB:-ON} WITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} WITH_INFRT=${WITH_INFRT:-OFF} INFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} PY_VERSION=${PY_VERSION:-3.7} CMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} WITH_PSCORE=${pscore_flag} WITH_PSLIB=${pslib_flag} WITH_GLOO=${gloo_flag} LITE_GIT_TAG=release/v2.10 WITH_XPU=${WITH_XPU:-OFF} WITH_MLU=${WITH_MLU:-OFF} WITH_IPU=${WITH_IPU:-OFF} WITH_CNCL=${WITH_CNCL:-OFF} XPU_SDK_ROOT=${XPU_SDK_ROOT:-""} WITH_LITE=${WITH_LITE:-OFF} WITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} WITH_ARM=${WITH_ARM:-OFF} WITH_ASCEND=${WITH_ASCEND:-OFF} WITH_ASCEND_CL=${WITH_ASCEND_CL:-OFF} WITH_ASCEND_INT64=${WITH_ASCEND_INT64:-OFF} WITH_STRIP=${WITH_STRIP:-ON} ON_INFER=${ON_INFER:-OFF} WITH_HETERPS=${WITH_HETERPS:-OFF} WITH_GPU_GRAPH=${WITH_GPU_GRAPH:-OFF} WITH_FLUID_ONLY=${WITH_FLUID_ONLY:-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=${WITH_CUDNN_FRONTEND:-OFF}
     ========================================
 EOF
     echo "if you use cmake to compile,please Configuring cmake in /paddle/build ..."
     cat <<EOF
     ========================================
-    cmake .. -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release} -DWITH_GPU=${WITH_GPU:-OFF} -DWITH_TENSORRT=${WITH_TENSORRT:-ON} -DWITH_ROCM=${WITH_ROCM:-OFF} -DWITH_CINN=${WITH_CINN:-OFF} -DWITH_DISTRIBUTE=${distibuted_flag} -DWITH_MKL=${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=${WITH_INCREMENTAL_COVERAGE:-OFF} -DCMAKE_MODULE_PATH=/opt/rocm/hip/cmake -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DWITH_CONTRIB=${WITH_CONTRIB:-ON} -DWITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} -DWITH_INFRT=${WITH_INFRT:-OFF} -DINFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} -DPY_VERSION=${PY_VERSION:-3.7} -DCMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} -DWITH_PSCORE=${pscore_flag} -DWITH_PSLIB=${pslib_flag} -DWITH_GLOO=${gloo_flag} -DLITE_GIT_TAG=release/v2.10 -DWITH_XPU=${WITH_XPU:-OFF} -DWITH_MLU=${WITH_MLU:-OFF} -DWITH_IPU=${WITH_IPU:-OFF} -DWITH_CNCL=${WITH_CNCL:-OFF} -DXPU_SDK_ROOT=${XPU_SDK_ROOT:-""} -DWITH_LITE=${WITH_LITE:-OFF} -DWITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} -DWITH_ARM=${WITH_ARM:-OFF} -DWITH_ASCEND=${WITH_ASCEND:-OFF} -DWITH_ASCEND_CL=${WITH_ASCEND_CL:-OFF} -DWITH_ASCEND_INT64=${WITH_ASCEND_INT64:-OFF} -DWITH_STRIP=${WITH_STRIP:-ON} -DON_INFER=${ON_INFER:-OFF} -DWITH_HETERPS=${WITH_HETERPS:-OFF} -DWITH_FLUID_ONLY=${WITH_FLUID_ONLY:-OFF} -DCUDA_ARCH_BIN=${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}
+    cmake .. -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release} -DWITH_GPU=${WITH_GPU:-OFF} -DWITH_TENSORRT=${WITH_TENSORRT:-ON} -DWITH_ROCM=${WITH_ROCM:-OFF} -DWITH_CINN=${WITH_CINN:-OFF} -DWITH_DISTRIBUTE=${distibuted_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_CONTRIB=${WITH_CONTRIB:-ON} -DWITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} -DWITH_INFRT=${WITH_INFRT:-OFF} -DINFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} -DPY_VERSION=${PY_VERSION:-3.7} -DCMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} -DWITH_PSCORE=${pscore_flag} -DWITH_PSLIB=${pslib_flag} -DWITH_GLOO=${gloo_flag} -DLITE_GIT_TAG=release/v2.10 -DWITH_XPU=${WITH_XPU:-OFF} -DWITH_MLU=${WITH_MLU:-OFF} -DWITH_IPU=${WITH_IPU:-OFF} -DWITH_CNCL=${WITH_CNCL:-OFF} -DXPU_SDK_ROOT=${XPU_SDK_ROOT:-""} -DWITH_LITE=${WITH_LITE:-OFF} -DWITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} -DWITH_ARM=${WITH_ARM:-OFF} -DWITH_ASCEND=${WITH_ASCEND:-OFF} -DWITH_ASCEND_CL=${WITH_ASCEND_CL:-OFF} -DWITH_ASCEND_INT64=${WITH_ASCEND_INT64:-OFF} -DWITH_STRIP=${WITH_STRIP:-ON} -DON_INFER=${ON_INFER:-OFF} -DWITH_HETERPS=${WITH_HETERPS:-OFF} -DWITH_GPU_GRAPH=${WITH_GPU_GRAPH:-OFF} -DWITH_FLUID_ONLY=${WITH_FLUID_ONLY:-OFF} -DCUDA_ARCH_BIN=${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}
     ========================================
 EOF
     export CMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release}
@@ -3691,6 +3693,7 @@ EOF
     export WITH_STRIP=${WITH_STRIP:-ON}
     export ON_INFER=${ON_INFER:-OFF}
     export WITH_HETERPS=${WITH_HETERPS:-OFF}
+    export WITH_GPU_GRAPH=${WITH_GPU_GRAPH:-OFF}
     export WITH_FLUID_ONLY=${WITH_FLUID_ONLY:-OFF}
     export CUDA_ARCH_BIN=${CUDA_ARCH_BIN}
     export WITH_RECORD_BUILDTIME=${WITH_RECORD_BUILDTIME:-OFF}
@@ -3886,13 +3889,13 @@ function run_setup_mac(){
     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_TENSORRT=${WITH_TENSORRT:-ON} WITH_ROCM=${WITH_ROCM:-OFF} WITH_CINN=${WITH_CINN:-OFF} WITH_DISTRIBUTE=${distibuted_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_CONTRIB=${WITH_CONTRIB:-ON} WITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} WITH_INFRT=${WITH_INFRT:-OFF} INFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} PY_VERSION=${PY_VERSION:-3.7} CMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} WITH_PSCORE=${pscore_flag} WITH_PSLIB=${pslib_flag} WITH_GLOO=${gloo_flag} LITE_GIT_TAG=release/v2.10 WITH_XPU=${WITH_XPU:-OFF} WITH_MLU=${WITH_MLU:-OFF} WITH_IPU=${WITH_IPU:-OFF} WITH_CNCL=${WITH_CNCL:-OFF} XPU_SDK_ROOT=${XPU_SDK_ROOT:-""} WITH_LITE=${WITH_LITE:-OFF} WITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} WITH_ARM=${WITH_ARM:-OFF} WITH_ASCEND=${WITH_ASCEND:-OFF} WITH_ASCEND_CL=${WITH_ASCEND_CL:-OFF} WITH_ASCEND_INT64=${WITH_ASCEND_INT64:-OFF} WITH_STRIP=${WITH_STRIP:-ON} ON_INFER=${ON_INFER:-OFF} WITH_HETERPS=${WITH_HETERPS:-OFF} WITH_FLUID_ONLY=${WITH_FLUID_ONLY:-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=${WITH_CUDNN_FRONTEND:-OFF}
+    export CMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release} WITH_GPU=${WITH_GPU:-OFF} WITH_TENSORRT=${WITH_TENSORRT:-ON} WITH_ROCM=${WITH_ROCM:-OFF} WITH_CINN=${WITH_CINN:-OFF} WITH_DISTRIBUTE=${distibuted_flag} WITH_MKL=${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=${WITH_INCREMENTAL_COVERAGE:-OFF} CMAKE_MODULE_PATH=/opt/rocm/hip/cmake CMAKE_EXPORT_COMPILE_COMMANDS=ON WITH_CONTRIB=${WITH_CONTRIB:-ON} WITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} WITH_INFRT=${WITH_INFRT:-OFF} INFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} PY_VERSION=${PY_VERSION:-3.7} CMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} WITH_PSCORE=${pscore_flag} WITH_PSLIB=${pslib_flag} WITH_GLOO=${gloo_flag} LITE_GIT_TAG=release/v2.10 WITH_XPU=${WITH_XPU:-OFF} WITH_MLU=${WITH_MLU:-OFF} WITH_IPU=${WITH_IPU:-OFF} WITH_CNCL=${WITH_CNCL:-OFF} XPU_SDK_ROOT=${XPU_SDK_ROOT:-""} WITH_LITE=${WITH_LITE:-OFF} WITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} WITH_ARM=${WITH_ARM:-OFF} WITH_ASCEND=${WITH_ASCEND:-OFF} WITH_ASCEND_CL=${WITH_ASCEND_CL:-OFF} WITH_ASCEND_INT64=${WITH_ASCEND_INT64:-OFF} WITH_STRIP=${WITH_STRIP:-ON} ON_INFER=${ON_INFER:-OFF} WITH_HETERPS=${WITH_HETERPS:-OFF} WITH_GPU_GRAPH=${WITH_GPU_GRAPH:-OFF} WITH_FLUID_ONLY=${WITH_FLUID_ONLY:-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=${WITH_CUDNN_FRONTEND:-OFF}
     ========================================
 EOF
     echo "if you use cmake to compile,please Configuring cmake in /paddle/build ..."
     cat <<EOF
     ========================================
-    cmake .. -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release} -DWITH_GPU=${WITH_GPU:-OFF} -DWITH_TENSORRT=${WITH_TENSORRT:-ON} -DWITH_ROCM=${WITH_ROCM:-OFF} -DWITH_CINN=${WITH_CINN:-OFF} -DWITH_DISTRIBUTE=${distibuted_flag} -DWITH_MKL=${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=${WITH_INCREMENTAL_COVERAGE:-OFF} -DCMAKE_MODULE_PATH=/opt/rocm/hip/cmake -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DWITH_CONTRIB=${WITH_CONTRIB:-ON} -DWITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} -DWITH_INFRT=${WITH_INFRT:-OFF} -DINFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} -DPY_VERSION=${PY_VERSION:-3.7} -DCMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} -DWITH_PSCORE=${pscore_flag} -DWITH_PSLIB=${pslib_flag} -DWITH_GLOO=${gloo_flag} -DLITE_GIT_TAG=release/v2.10 -DWITH_XPU=${WITH_XPU:-OFF} -DWITH_MLU=${WITH_MLU:-OFF} -DWITH_IPU=${WITH_IPU:-OFF} -DWITH_CNCL=${WITH_CNCL:-OFF} -DXPU_SDK_ROOT=${XPU_SDK_ROOT:-""} -DWITH_LITE=${WITH_LITE:-OFF} -DWITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} -DWITH_ARM=${WITH_ARM:-OFF} -DWITH_ASCEND=${WITH_ASCEND:-OFF} -DWITH_ASCEND_CL=${WITH_ASCEND_CL:-OFF} -DWITH_ASCEND_INT64=${WITH_ASCEND_INT64:-OFF} -DWITH_STRIP=${WITH_STRIP:-ON} -DON_INFER=${ON_INFER:-OFF} -DWITH_HETERPS=${WITH_HETERPS:-OFF} -DWITH_FLUID_ONLY=${WITH_FLUID_ONLY:-OFF} -DCUDA_ARCH_BIN=${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}
+    cmake .. -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release} -DWITH_GPU=${WITH_GPU:-OFF} -DWITH_TENSORRT=${WITH_TENSORRT:-ON} -DWITH_ROCM=${WITH_ROCM:-OFF} -DWITH_CINN=${WITH_CINN:-OFF} -DWITH_DISTRIBUTE=${distibuted_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_CONTRIB=${WITH_CONTRIB:-ON} -DWITH_INFERENCE_API_TEST=${WITH_INFERENCE_API_TEST:-ON} -DWITH_INFRT=${WITH_INFRT:-OFF} -DINFERENCE_DEMO_INSTALL_DIR=${INFERENCE_DEMO_INSTALL_DIR} -DPY_VERSION=${PY_VERSION:-3.7} -DCMAKE_INSTALL_PREFIX=${INSTALL_PREFIX:-/paddle/build} -DWITH_PSCORE=${pscore_flag} -DWITH_PSLIB=${pslib_flag} -DWITH_GLOO=${gloo_flag} -DLITE_GIT_TAG=release/v2.10 -DWITH_XPU=${WITH_XPU:-OFF} -DWITH_MLU=${WITH_MLU:-OFF} -DWITH_IPU=${WITH_IPU:-OFF} -DWITH_CNCL=${WITH_CNCL:-OFF} -DXPU_SDK_ROOT=${XPU_SDK_ROOT:-""} -DWITH_LITE=${WITH_LITE:-OFF} -DWITH_XPU_BKCL=${WITH_XPU_BKCL:-OFF} -DWITH_ARM=${WITH_ARM:-OFF} -DWITH_ASCEND=${WITH_ASCEND:-OFF} -DWITH_ASCEND_CL=${WITH_ASCEND_CL:-OFF} -DWITH_ASCEND_INT64=${WITH_ASCEND_INT64:-OFF} -DWITH_STRIP=${WITH_STRIP:-ON} -DON_INFER=${ON_INFER:-OFF} -DWITH_HETERPS=${WITH_HETERPS:-OFF} -DWITH_GPU_GRAPH=${WITH_GPU_GRAPH:-OFF} -DWITH_FLUID_ONLY=${WITH_FLUID_ONLY:-OFF} -DCUDA_ARCH_BIN=${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}
     ========================================
 EOF
     export CMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE:-Release}
@@ -3939,6 +3942,7 @@ EOF
     export WITH_STRIP=${WITH_STRIP:-ON}
     export ON_INFER=${ON_INFER:-OFF}
     export WITH_HETERPS=${WITH_HETERPS:-OFF}
+    export WITH_GPU_GRAPH=${WITH_GPU_GRAPH:-OFF}
     export WITH_FLUID_ONLY=${WITH_FLUID_ONLY:-OFF}
     export CUDA_ARCH_BIN=${CUDA_ARCH_BIN}
     export WITH_RECORD_BUILDTIME=${WITH_RECORD_BUILDTIME:-OFF}