Merge branch 'PaddlePaddle:develop' into fix_trt_int8_infer

.clang-tidy

@@ -140,7 +140,7 @@ clang-analyzer-optin.portability.UnixAPI,
 clang-analyzer-security.insecureAPI.vfork,
 -clang-analyzer-unix.API,
 -clang-analyzer-unix.DynamicMemoryModeling,
--clang-analyzer-unix.Malloc,
+clang-analyzer-unix.Malloc,
 -clang-analyzer-unix.MallocSizeof,
 -clang-analyzer-unix.MismatchedDeallocator,
 clang-analyzer-unix.Vfork,
@@ -158,7 +158,7 @@ cppcoreguidelines-explicit-virtual-functions,
 cppcoreguidelines-init-variables,
 cppcoreguidelines-narrowing-conversions,
 cppcoreguidelines-no-malloc,
--cppcoreguidelines-pro-type-const-cast,
+cppcoreguidelines-pro-type-const-cast,
 -cppcoreguidelines-pro-type-member-init,
 -cppcoreguidelines-slicing,
 -hicpp-avoid-goto,

.gitignore

@@ -100,3 +100,9 @@ paddle/fluid/ir_adaptor/translator/op_compat_info.cc
 paddle/phi/kernels/fusion/cutlass/cutlass_kernels/fpA_intB_gemm/autogen/*
 paddle/fluid/pybind/static_op_function.*
 paddle/fluid/pybind/ops_api.cc
+paddle/fluid/pir/dialect/operator/ir/pd_api.*
+paddle/fluid/pir/dialect/operator/ir/op_decomp.cc
+paddle/fluid/pir/dialect/operator/ir/pd_op_vjp.cc
+paddle/fluid/pir/dialect/operator/ir/pd_op.*
+paddle/cinn/hlir/dialect/generated/ops.parsed.yaml
+paddle/cinn/hlir/dialect/operator/ir/cinn_op.*

.pre-commit-config.yaml

@@ -56,28 +56,16 @@ repos:
     hooks:
     -   id: black
         files: (.*\.(py|pyi|bzl)|BUILD|.*\.BUILD|WORKSPACE)$
--   repo: https://github.com/pycqa/isort
-    rev: 5.11.5
-    hooks:
-    -   id: isort
 -   repo: https://github.com/PyCQA/flake8
     rev: 5.0.4
     hooks:
     -   id: flake8
         args: ["--config=.flake8"]
 -   repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.0.289
+    rev: v0.1.5
     hooks:
     -   id: ruff
         args: [--fix, --exit-non-zero-on-fix, --no-cache]
--   repo: local
-    hooks:
-    -   id: pylint-doc-string
-        name: pylint
-        description: Check python docstring style using docstring_checker.
-        entry: bash ./tools/codestyle/pylint_pre_commit.hook
-        language: system
-        files: \.(py)$
 # For C++ files
 -   repo: local
     hooks:
@@ -155,18 +143,18 @@ repos:
             (?x)^(
                 paddle/fluid/operators/CMakeLists.txt
             )$
--   repo: https://github.com/cmake-lint/cmake-lint
-    rev: 1.4.2
+-   repo: https://github.com/PFCCLab/cmake-lint-paddle
+    rev: v1.5.1
     hooks:
     -   id: cmakelint
         args: [--config=./tools/codestyle/.cmakelintrc]
-
-- repo: local
-  hooks:
-    - id: sort-txt-file
-      name: sort-txt-file
-      description: Sorts each line string in a text file
-      entry: python ./tools/codestyle/sort_txt_file.py
-      language: python
-      files: test/white_list/new_ir_op_test_white_list
-      args: []
+# Others
+-   repo: local
+    hooks:
+    -   id: sort-txt-file
+        name: sort-txt-file
+        description: Sorts each line string in a text file
+        entry: python ./tools/codestyle/sort_txt_file.py
+        language: python
+        files: test/white_list/pir_op_test_white_list
+        args: []

cmake/cinn.cmake

@@ -70,6 +70,10 @@ if(WITH_GPU)
     message(STATUS "Enable CINN CUDNN")
     add_definitions(-DCINN_WITH_CUDNN)
   endif()
+  if(WITH_CUTLASS)
+    message(STATUS "Enable CINN CUTLASS")
+    add_definitions(-DCINN_WITH_CUTLASS)
+  endif()
   enable_language(CUDA)
   find_package(CUDA REQUIRED)
   include_directories(${CUDA_INCLUDE_DIRS})
@@ -164,8 +168,8 @@ cinn_cc_library(
 add_dependencies(cinnapi GEN_LLVM_RUNTIME_IR_HEADER ZLIB::ZLIB)
 add_dependencies(cinnapi GEN_LLVM_RUNTIME_IR_HEADER ${core_deps})
 if(NOT CINN_ONLY)
-  target_link_libraries(cinnapi pd_op_dialect phi)
-  add_dependencies(cinnapi pd_op_dialect phi)
+  target_link_libraries(cinnapi op_dialect_vjp phi)
+  add_dependencies(cinnapi op_dialect_vjp phi)
 endif()
 
 target_link_libraries(cinnapi ${PYTHON_LIBRARIES})
@@ -199,6 +203,11 @@ if(WITH_GPU)
   endif()
 endif()
 
+if(WITH_CUTLASS)
+  target_link_libraries(cinnapi cutlass)
+  add_dependencies(cinnapi cutlass)
+endif()
+
 function(gen_cinncore LINKTYPE)
   set(CINNCORE_TARGET cinncore)
   if(${LINKTYPE} STREQUAL "STATIC")
@@ -222,8 +231,8 @@ function(gen_cinncore LINKTYPE)
   add_dependencies(${CINNCORE_TARGET} GEN_LLVM_RUNTIME_IR_HEADER ZLIB::ZLIB)
   add_dependencies(${CINNCORE_TARGET} GEN_LLVM_RUNTIME_IR_HEADER ${core_deps})
   if(NOT CINN_ONLY)
-    target_link_libraries(${CINNCORE_TARGET} pd_op_dialect phi)
-    add_dependencies(${CINNCORE_TARGET} pd_op_dialect phi)
+    target_link_libraries(${CINNCORE_TARGET} op_dialect_vjp phi)
+    add_dependencies(${CINNCORE_TARGET} op_dialect_vjp phi)
   endif()
 
   add_dependencies(${CINNCORE_TARGET} pybind)
@@ -258,6 +267,11 @@ function(gen_cinncore LINKTYPE)
       target_link_libraries(${CINNCORE_TARGET} ${CUDA_NVTX_LIB})
     endif()
   endif()
+
+  if(WITH_CUTLASS)
+    target_link_libraries(cinnapi cutlass)
+    add_dependencies(cinnapi cutlass)
+  endif()
 endfunction()
 
 gen_cinncore(STATIC)

cmake/external/cccl.cmake

@@ -26,6 +26,7 @@ ExternalProject_Add(
   INSTALL_COMMAND ""
   TEST_COMMAND "")
 
-add_library(cccl INTERFACE)
-
-add_dependencies(cccl extern_cccl)
+# update include dir and set cccl first for using
+include_directories(BEFORE "${CCCL_SOURCE_DIR}/cub")
+include_directories(BEFORE "${CCCL_SOURCE_DIR}/libcudacxx/include")
+include_directories(BEFORE "${CCCL_SOURCE_DIR}/thrust")

cmake/external/gloo.cmake

@@ -61,6 +61,17 @@ if(CMAKE_COMPILER_IS_GNUCC)
         ${SOURCE_DIR}/gloo/ < ${types_header})
   endif()
 endif()
+
+file(TO_NATIVE_PATH ${PADDLE_SOURCE_DIR}/patches/gloo/linux.cc.patch
+     linux_cc_ethtool)
+if(GLOO_PATCH_COMMAND STREQUAL "")
+  set(GLOO_PATCH_COMMAND git checkout -- . && git checkout ${GLOO_TAG} && patch
+                         -Nd ${SOURCE_DIR}/gloo/common/ < ${linux_cc_ethtool})
+else()
+  set(GLOO_PATCH_COMMAND ${GLOO_PATCH_COMMAND} && patch -Nd
+                         ${SOURCE_DIR}/gloo/common/ < ${linux_cc_ethtool})
+endif()
+
 include_directories(${GLOO_INCLUDE_DIR})
 
 ExternalProject_Add(

cmake/external/protobuf.cmake

@@ -182,7 +182,7 @@ if(NOT "${PROTOBUF_ROOT}" STREQUAL "")
   endif()
 endif()
 
-function(build_protobuf TARGET_NAME BUILD_FOR_HOST)
+function(build_protobuf TARGET_NAME)
   string(REPLACE "extern_" "" TARGET_DIR_NAME "${TARGET_NAME}")
   set(PROTOBUF_PREFIX_DIR ${THIRD_PARTY_PATH}/${TARGET_DIR_NAME})
   set(PROTOBUF_SOURCE_DIR
@@ -210,22 +210,18 @@ function(build_protobuf TARGET_NAME BUILD_FOR_HOST)
       PARENT_SCOPE)
 
   set(OPTIONAL_CACHE_ARGS "")
-  set(OPTIONAL_ARGS "")
-  if(BUILD_FOR_HOST)
-    set(OPTIONAL_ARGS "-Dprotobuf_WITH_ZLIB=OFF")
-  else()
-    set(OPTIONAL_ARGS
-        "-DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}"
-        "-DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}"
-        "-DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}"
-        "-DCMAKE_C_FLAGS_DEBUG=${CMAKE_C_FLAGS_DEBUG}"
-        "-DCMAKE_C_FLAGS_RELEASE=${CMAKE_C_FLAGS_RELEASE}"
-        "-DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}"
-        "-DCMAKE_CXX_FLAGS_RELEASE=${CMAKE_CXX_FLAGS_RELEASE}"
-        "-DCMAKE_CXX_FLAGS_DEBUG=${CMAKE_CXX_FLAGS_DEBUG}"
-        "-Dprotobuf_WITH_ZLIB=ON"
-        "-DZLIB_ROOT:FILEPATH=${ZLIB_ROOT}"
-        ${EXTERNAL_OPTIONAL_ARGS})
+  set(OPTIONAL_ARGS
+      "-DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}"
+      "-DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}"
+      "-DCMAKE_C_FLAGS=${CMAKE_C_FLAGS}"
+      "-DCMAKE_C_FLAGS_DEBUG=${CMAKE_C_FLAGS_DEBUG}"
+      "-DCMAKE_C_FLAGS_RELEASE=${CMAKE_C_FLAGS_RELEASE}"
+      "-DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}"
+      "-DCMAKE_CXX_FLAGS_RELEASE=${CMAKE_CXX_FLAGS_RELEASE}"
+      "-DCMAKE_CXX_FLAGS_DEBUG=${CMAKE_CXX_FLAGS_DEBUG}"
+      "-Dprotobuf_WITH_ZLIB=ON"
+      ${EXTERNAL_OPTIONAL_ARGS})
+  if(NOT APPLE)
     set(OPTIONAL_CACHE_ARGS "-DZLIB_ROOT:STRING=${ZLIB_ROOT}")
   endif()
   if(WIN32)
@@ -339,7 +335,7 @@ else()
 endif()
 
 if(NOT PROTOBUF_FOUND)
-  build_protobuf(extern_protobuf FALSE)
+  build_protobuf(extern_protobuf)
 
   set(PROTOBUF_INCLUDE_DIR
       ${extern_protobuf_INCLUDE_DIR}

cmake/external/xpu.cmake

@@ -24,9 +24,9 @@ set(XPU_XFT_LIB_NAME "libxft.so")
 set(XPU_XPTI_LIB_NAME "libxpti.so")
 
 if(NOT DEFINED XPU_BASE_DATE)
-  set(XPU_BASE_DATE "20231025")
+  set(XPU_BASE_DATE "20231114")
 endif()
-set(XPU_XCCL_BASE_VERSION "1.0.53.6")
+set(XPU_XCCL_BASE_VERSION "1.1.6.1")
 if(NOT DEFINED XPU_XFT_BASE_VERSION)
   set(XPU_XFT_BASE_VERSION "20230602")
 endif()

cmake/generic.cmake

@@ -90,7 +90,6 @@
 #
 #   paddle_test(example SRCS example_test.cc)
 #
-
 # including binary directory for generated headers.
 include_directories(${CMAKE_CURRENT_BINARY_DIR})
 # including io directory for inference lib paddle_api.h
@@ -1349,9 +1348,6 @@ function(math_library TARGET)
   if(WITH_GPU)
     if(${CMAKE_CUDA_COMPILER_VERSION} LESS 11.0)
       list(APPEND math_common_deps cub)
-    elseif(${CMAKE_CUDA_COMPILER_VERSION} EQUAL 12.0
-           OR ${CMAKE_CUDA_COMPILER_VERSION} GREATER 12.0)
-      list(APPEND math_common_deps cccl)
     else()
       list(APPEND math_common_deps)
     endif()

cmake/inference_lib.cmake

@@ -268,13 +268,11 @@ else()
       SRCS ${paddle_phi_lib}
       DSTS ${PADDLE_INFERENCE_INSTALL_DIR}/paddle/lib)
   endif()
-  if(WITH_SHARED_IR)
-    set(paddle_pir_lib ${PADDLE_BINARY_DIR}/paddle/pir/libpir.*)
-    copy(
-      inference_lib_dist
-      SRCS ${paddle_pir_lib}
-      DSTS ${PADDLE_INFERENCE_INSTALL_DIR}/paddle/lib)
-  endif()
+  set(paddle_common_lib ${PADDLE_BINARY_DIR}/paddle/common/libcommon.*)
+  copy(
+    inference_lib_dist
+    SRCS ${paddle_common_lib}
+    DSTS ${PADDLE_INFERENCE_INSTALL_DIR}/paddle/lib)
 endif()
 
 copy(

cmake/third_party.cmake

@@ -272,6 +272,17 @@ include(external/gflags) # download, build, install gflags
 include(external/glog) # download, build, install glog
 
 ########################### include third_party according to flags ###############################
+if(WITH_GPU
+   AND NOT WITH_ARM
+   AND NOT WIN32
+   AND NOT APPLE)
+  if(${CMAKE_CUDA_COMPILER_VERSION} GREATER_EQUAL 11.0)
+    include(external/cutlass) # download, build, install cusparselt
+    list(APPEND third_party_deps extern_cutlass)
+    set(WITH_CUTLASS ON)
+  endif()
+endif()
+
 if(WITH_CINN)
   if(WITH_MKL)
     add_definitions(-DCINN_WITH_MKL_CBLAS)
@@ -386,7 +397,7 @@ if(WITH_GPU)
   elseif(${CMAKE_CUDA_COMPILER_VERSION} EQUAL 12.0
          OR ${CMAKE_CUDA_COMPILER_VERSION} GREATER 12.0)
     include(external/cccl)
-    list(APPEND third_party_deps extern_cccl)
+    add_definitions(-DPADDLE_WITH_CCCL)
   endif()
   set(URL
       "https://paddlepaddledeps.bj.bcebos.com/externalErrorMsg_20210928.tar.gz"
@@ -555,11 +566,6 @@ if(WITH_GPU
    AND NOT WITH_ARM
    AND NOT WIN32
    AND NOT APPLE)
-  if(${CMAKE_CUDA_COMPILER_VERSION} GREATER_EQUAL 11.0)
-    include(external/cutlass) # download, build, install cusparselt
-    list(APPEND third_party_deps extern_cutlass)
-    set(WITH_CUTLASS ON)
-  endif()
   if(${CMAKE_CUDA_COMPILER_VERSION} GREATER_EQUAL 11.4)
     foreach(arch ${NVCC_ARCH_BIN})
       if(${arch} GREATER_EQUAL 80)

paddle/CMakeLists.txt

@@ -3,6 +3,7 @@ set(PYTHON_TESTS_DIR
     CACHE INTERNAL "python tests directory")
 
 add_subdirectory(utils)
+add_subdirectory(common)
 add_subdirectory(pir)
 add_subdirectory(scripts)
 add_subdirectory(testing)

paddle/cinn/CMakeLists.txt

@@ -2,6 +2,7 @@ if(WITH_TESTING)
   cinn_cc_library(cinn_gtest_main SRCS gtest_main.cc DEPS gtest ${flags_dep})
 endif()
 
+add_subdirectory(adt)
 add_subdirectory(api)
 add_subdirectory(ast_gen_ius)
 add_subdirectory(auto_schedule)

paddle/cinn/adt/CMakeLists.txt

@@ -0,0 +1,39 @@
+if(NOT CINN_ONLY)
+  add_subdirectory(print_utils)
+
+  core_gather_headers()
+
+  gather_srcs(
+    cinnapi_src
+    SRCS
+    adapter_tensor.cc
+    anchor_sd_equation_context.cc
+    equation_function.cc
+    equation_solver.cc
+    equation_value.cc
+    generate_map_expr.cc
+    get_sub_reshape_dim_ranges.cc
+    igroup.cc
+    index_expr_infer_context.cc
+    kgroup.cc
+    m_ir.cc
+    naive_bidirection_equation_generator.cc
+    naive_op_equation_context.cc
+    partition_op_stmts.cc
+    schedule_descriptor.cc
+    schedule_dim.cc
+    schedule_mesh.cc
+    simplify_value.cc
+    write_broadcast_disabled_bidirection_equation_generator.cc)
+
+  cinn_cc_test(equation_value_match_trait_test SRCS
+               equation_value_match_trait_test.cc DEPS gtest glog)
+
+  cinn_cc_test(tree_test SRCS tree_test.cc DEPS gtest glog)
+
+  cinn_cc_test(inline_translator_test SRCS inline_translator_test.cc DEPS
+               cinncore)
+
+  message(STATUS "ADT srcs: ${cinnapi_src}")
+
+endif()