Merge remote-tracking branch 'upstream/develop' into pow-cp

Author: zbt78

.github/PULL_REQUEST_TEMPLATE.md

@@ -1,9 +1,13 @@
+<!-- TemplateReference: https://github.com/PaddlePaddle/Paddle/wiki/PULL-REQUEST-TEMPLATE--REFERENCE -->
 <!-- Demo: https://github.com/PaddlePaddle/Paddle/pull/24810 -->
-### PR types
-<!-- One of [ New features | Bug fixes | Function optimization | Performance optimization | Breaking changes | Others ] -->
 
-### PR changes
-<!-- One of [ OPs | APIs | Docs | Others ] -->
+### PR Category
+<!-- One of [ User Experience | Execute Infrastructure | Operator Mechanism | CINN | Custom Device | Performance Optimization | Distributed Strategy | Parameter Server | Communication Library | Auto Parallel | Inference | Environment Adaptation | Others ] -->
+
+
+### PR Types
+<!-- One of [ New features | Bug fixes | Improvements | Performance | BC Breaking | Deprecations | Docs | Devs | Not User Facing | Security | Deprecations | Others ] -->
+
 
 ### Description
 <!-- Describe what you’ve done -->

.gitmodules

@@ -70,8 +70,8 @@
 	path = third_party/xbyak
 	url = https://github.com/herumi/xbyak.git
 	ignore = dirty
-[submodule "third_party/mkldnn"]
-	path = third_party/mkldnn
+[submodule "third_party/onednn"]
+	path = third_party/onednn
 	url = https://github.com/oneapi-src/oneDNN.git
 	ignore = dirty
 [submodule "third_party/flashattn"]
@@ -118,3 +118,7 @@
 	path = third_party/cryptopp-cmake
 	url = https://github.com/noloader/cryptopp-cmake.git
 	ignore = dirty
+[submodule "third_party/nlohmann_json"]
+	path = third_party/nlohmann_json
+	url = https://github.com/nlohmann/json.git
+	ignore = dirty

.pre-commit-config.yaml

@@ -42,7 +42,7 @@ repos:
     hooks:
     -   id: copyright_checker
         name: copyright_checker
-        entry: python ./tools/codestyle/copyright.hook
+        entry: python ./tools/codestyle/copyright.py
         language: system
         files: \.(c|cc|cxx|cpp|cu|h|hpp|hxx|proto|xpu|kps|py|sh)$
         exclude: |
@@ -57,7 +57,7 @@ repos:
     -   id: black
         files: (.*\.(py|pyi|bzl)|BUILD|.*\.BUILD|WORKSPACE)$
 -   repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.3.0
+    rev: v0.3.5
     hooks:
     -   id: ruff
         args: [--fix, --exit-non-zero-on-fix, --no-cache]
@@ -67,15 +67,15 @@ repos:
     -   id: clang-format
         name: clang-format
         description: Format files with ClangFormat.
-        entry: bash ./tools/codestyle/clang_format.hook -i
+        entry: bash ./tools/codestyle/clang_format.sh -i
         language: system
         files: \.(c|cc|cxx|cpp|cu|h|hpp|hxx|xpu|kps)$
 -   repo: local
     hooks:
     -   id: cpplint-cpp-source
         name: cpplint
         description: Check C++ code style using cpplint.py.
-        entry: bash ./tools/codestyle/cpplint_pre_commit.hook
+        entry: bash ./tools/codestyle/cpplint_pre_commit.sh
         language: system
         files: \.(cc|cxx|cpp|cu|h|hpp|hxx)$
         args:

CMakeLists.txt

@@ -142,7 +142,10 @@ endif()
 if(WIN32)
   option(MSVC_STATIC_CRT "use static C Runtime library by default" ON)
   message("Build static library of PHI")
-  set(CMAKE_SUPPRESS_REGENERATION ON)
+  # (Note xuxinyi04): If CMAKE_SUPPRESS_REGENERATION is OFF, which is default, then CMake adds a
+  # special target on which all other targets depend that checks the build system and optionally
+  # re-runs CMake to regenerate the build system when the target specification source changes.
+  set(CMAKE_SUPPRESS_REGENERATION OFF)
   set(CMAKE_STATIC_LIBRARY_PREFIX lib)
   set(WITH_SHARED_PHI
       OFF
@@ -303,7 +306,6 @@ option(
   SANITIZER_TYPE
   "Choose the type of sanitizer, options are: Address, Leak, Memory, Thread, Undefined"
   OFF)
-option(WITH_LITE "Compile Paddle Fluid with Lite Engine" OFF)
 option(WITH_CINN "Compile PaddlePaddle with CINN" OFF)
 option(WITH_NCCL "Compile PaddlePaddle with NCCL support" ON)
 option(WITH_RCCL "Compile PaddlePaddle with RCCL support" ON)

ISSUE_TEMPLATE.md

@@ -59,7 +59,7 @@ if(WITH_MKL)
   add_dependencies(cinn_mklml ${MKLML_PROJECT})
   add_definitions(-DCINN_WITH_MKL_CBLAS)
 endif()
-if(WITH_MKLDNN)
+if(WITH_ONEDNN)
   add_definitions(-DCINN_WITH_DNNL)
 endif()
 
@@ -164,6 +164,8 @@ cinn_cc_library(
   isl
   ginac
   pybind
+  op_fusion
+  cinn_op_dialect
   ${jitify_deps})
 add_dependencies(cinnapi GEN_LLVM_RUNTIME_IR_HEADER ZLIB::ZLIB)
 add_dependencies(cinnapi GEN_LLVM_RUNTIME_IR_HEADER ${core_deps})
@@ -175,9 +177,9 @@ target_link_libraries(cinnapi ${PYTHON_LIBRARIES})
 if(WITH_MKL)
   target_link_libraries(cinnapi cinn_mklml)
   add_dependencies(cinnapi cinn_mklml)
-  if(WITH_MKLDNN)
-    target_link_libraries(cinnapi ${MKLDNN_LIB})
-    add_dependencies(cinnapi ${MKLDNN_PROJECT})
+  if(WITH_ONEDNN)
+    target_link_libraries(cinnapi ${ONEDNN_LIB})
+    add_dependencies(cinnapi ${ONEDNN_PROJECT})
   endif()
 endif()
 
@@ -220,21 +222,25 @@ function(gen_cinncore LINKTYPE)
     schedule_desc_proto
     absl
     isl
-    ginac)
+    ginac
+    pybind
+    op_fusion
+    cinn_op_dialect
+    ${jitify_deps})
   add_dependencies(${CINNCORE_TARGET} GEN_LLVM_RUNTIME_IR_HEADER ZLIB::ZLIB)
   add_dependencies(${CINNCORE_TARGET} GEN_LLVM_RUNTIME_IR_HEADER ${core_deps})
   target_link_libraries(${CINNCORE_TARGET} op_dialect pir phi)
   add_dependencies(${CINNCORE_TARGET} op_dialect pir phi)
 
-  add_dependencies(${CINNCORE_TARGET} pybind)
+  # add_dependencies(${CINNCORE_TARGET} pybind)
   target_link_libraries(${CINNCORE_TARGET} ${PYTHON_LIBRARIES})
 
   if(WITH_MKL)
     target_link_libraries(${CINNCORE_TARGET} cinn_mklml)
     add_dependencies(${CINNCORE_TARGET} cinn_mklml)
-    if(WITH_MKLDNN)
-      target_link_libraries(${CINNCORE_TARGET} ${MKLDNN_LIB})
-      add_dependencies(${CINNCORE_TARGET} ${MKLDNN_PROJECT})
+    if(WITH_ONEDNN)
+      target_link_libraries(${CINNCORE_TARGET} ${ONEDNN_LIB})
+      add_dependencies(${CINNCORE_TARGET} ${ONEDNN_PROJECT})
     endif()
   endif()
 
@@ -247,16 +253,16 @@ function(gen_cinncore LINKTYPE)
       ${CUBLAS}
       ${CUDNN}
       ${CURAND}
-      ${CUSOLVER}
-      ${jitify_deps})
+      ${CUSOLVER})
+    # ${jitify_deps})
     if(NVTX_FOUND)
       target_link_libraries(${CINNCORE_TARGET} ${CUDA_NVTX_LIB})
     endif()
   endif()
 
   if(WITH_CUTLASS)
-    target_link_libraries(cinnapi cutlass)
-    add_dependencies(cinnapi cutlass)
+    target_link_libraries(${CINNCORE_TARGET} cutlass)
+    add_dependencies(${CINNCORE_TARGET} cutlass)
   endif()
 endfunction()
 

cmake/cinn.cmake

@@ -60,8 +60,8 @@ endfunction()
 
 if(WITH_COVERAGE)
   if(WITH_INCREMENTAL_COVERAGE)
-    # if *.h changed, generate coverage report totaly.
-    # if pybind.cc changed, generate coverage report totaly.
+    # if *.h changed, generate coverage report totally.
+    # if pybind.cc changed, generate coverage report totally.
     # Because if pybind.cc add '-g -O0 -fprofile-arcs -ftest-coverage' only, some testcase will fail.
     if((NOT ("$ENV{PADDLE_GIT_DIFF_H_FILE}" STREQUAL ""))
        OR ("$ENV{PADDLE_GIT_DIFF_CC_FILE}" MATCHES "pybind.cc"))

cmake/coveralls.cmake

@@ -248,7 +248,7 @@ foreach(GCOV_FILE ${GCOV_FILES})
   # Instead of trying to parse the source from the
   # gcov file, simply read the file contents from the source file.
   # (Parsing it from the gcov is hard because C-code uses ; in many places
-  #  which also happens to be the same as the CMake list delimeter).
+  #  which also happens to be the same as the CMake list delimiter).
   file(READ ${GCOV_SRC_PATH} GCOV_FILE_SOURCE)
 
   string(REPLACE "\\" "\\\\" GCOV_FILE_SOURCE "${GCOV_FILE_SOURCE}")

cmake/coverallsGcovJsons.cmake

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 # this file contains experimental build options for lazy cuda module loading
-# cuda moduel lazy loading is supported by CUDA 11.7+
+# cuda module lazy loading is supported by CUDA 11.7+
 # this experiment option makes Paddle supports lazy loading before CUDA 11.7.
 
 if(LINUX)

cmake/experiments/cuda_module_loading_lazy.cmake

@@ -25,7 +25,7 @@ if(WIN32)
 elseif(LINUX)
   if(WITH_ROCM)
     # For HIPCC Eigen::internal::device::numeric_limits is not EIGEN_DEVICE_FUNC
-    # which will cause compiler error of using __host__ funciont
+    # which will cause compiler error of using __host__ function
     # in __host__ __device__
     file(TO_NATIVE_PATH ${PADDLE_SOURCE_DIR}/patches/eigen/Meta.h native_src)
     file(TO_NATIVE_PATH ${SOURCE_DIR}/Eigen/src/Core/util/Meta.h native_dst)