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

Author: caolonghao

.flake8

@@ -1,22 +1,95 @@
 [flake8]
 select = C,E,F,W
-exclude = ./build
+exclude =
+    ./build,
+    # A trick to exclude fluid/ but keep fluid/tests/, see more at
+    # https://github.com/PaddlePaddle/Paddle/pull/46290#discussion_r976392010
+    ./python/paddle/fluid/[!t]**,
+    ./python/paddle/fluid/tra**
 ignore =
     # E, see https://pycodestyle.pycqa.org/en/latest/intro.html#error-codes
-    E101,E121,E122,E123,E124,E125,E126,E127,E128,E129,E131,
-    E201,E202,E203,E221,E225,E226,E228,E231,E241,E251,E261,E262,E265,E266,E271,E272,E275,
+    E121,E122,E123,E125,E126,E127,E128,E129,E131,
+    E201,E202,E203,E225,E226,E228,E231,E241,E251,E261,E262,E265,E266,E271,E272,
     E301,E302,E303,E305,E306,
     E401,E402,
     E501,E502,
     E701,E711,E712,E713,E714,E721,E722,E731,E741,
 
     # F, see https://flake8.pycqa.org/en/latest/user/error-codes.html
-    F401,F402,F403,F404,F405,
-    F522,F523,F524,F541,
+    F402,F403,F405,
+    F522,F524,F541,
     F601,F631,F632,
-    F811,F821,F823,F841,
+    F811,F821,F841,
 
     # W, see https://pycodestyle.pycqa.org/en/latest/intro.html#error-codes
-    W191,
     W503,W504
     W601,W605
+per-file-ignores =
+    # These files need tabs for testing.
+    python/paddle/fluid/tests/unittests/dygraph_to_static/test_error.py:E101,W191
+    python/paddle/fluid/tests/unittests/collective/fleet/test_hdfs1.py:E101,W191
+    # Ignore unused imports in __init__.py
+    __init__.py: F401
+    # These files will be fixed later
+    r/*:F401
+    cmake/*:F401
+    paddle/*:F401
+    tools/*:F401
+    python/paddle/signal.py:F401
+    python/paddle/common_ops_import.py:F401
+    python/paddle/check_import_scipy.py:F401
+    python/paddle/fft.py:F401
+    python/paddle/_C_ops.py:F401
+    python/paddle/framework/*:F401
+    python/paddle/reader/*:F401
+    python/paddle/nn/*:F401
+    python/paddle/distributed/*:F401
+    python/paddle/proto/*:F401
+    python/paddle/onnx/*:F401
+    python/paddle/optimizer/*:F401
+    python/paddle/hapi/*:F401
+    python/paddle/autograd/*:F401
+    python/paddle/dataset/*:F401
+    python/paddle/io/*:F401
+    python/paddle/cost_model/*:F401
+    python/paddle/tests/*:F401
+    python/paddle/distribution/*:F401
+    python/paddle/geometric/*:F401
+    python/paddle/utils/*:F401
+    python/paddle/vision/*:F401
+    python/paddle/quantization/*:F401
+    python/paddle/libs/*:F401
+    python/paddle/audio/*:F401
+    python/paddle/incubate/*:F401
+    python/paddle/amp/*:F401
+    python/paddle/jit/*:F401
+    python/paddle/static/*:F401
+    python/paddle/inference/*:F401
+    python/paddle/device/*:F401
+    python/paddle/profiler/*:F401
+    python/paddle/tensor/*:F401
+    python/paddle/text/*:F401
+    python/paddle/metric/*:F401
+    python/paddle/fluid/tests/custom_kernel/*:F401
+    python/paddle/fluid/tests/custom_runtime/*:F401
+    python/paddle/fluid/tests/unittests/ir/*:F401
+    python/paddle/fluid/tests/unittests/tokenizer/*:F401
+    python/paddle/fluid/tests/unittests/xpu/*:F401
+    python/paddle/fluid/tests/unittests/distribution/*:F401
+    python/paddle/fluid/tests/unittests/mlu/*:F401
+    python/paddle/fluid/tests/unittests/npu/*:F401
+    python/paddle/fluid/tests/unittests/ipu/*:F401
+    python/paddle/fluid/tests/unittests/distributed_passes/*:F401
+    python/paddle/fluid/tests/unittests/auto_parallel/*:F401
+    python/paddle/fluid/tests/unittests/dygraph_to_static/*:F401
+    python/paddle/fluid/tests/unittests/ps/*:F401
+    python/paddle/fluid/tests/unittests/fft/*:F401
+    python/paddle/fluid/tests/unittests/white_list/*:F401
+    python/paddle/fluid/tests/unittests/sequence/*:F401
+    python/paddle/fluid/tests/unittests/mkldnn/*:F401
+    python/paddle/fluid/tests/unittests/rnn/*:F401
+    python/paddle/fluid/tests/book/*:F401
+    python/paddle/fluid/tests/custom_op/*:F401
+    python/paddle/fluid/tests/unittests/test_*:F401
+    python/paddle/fluid/tests/test_*:F401
+    python/paddle/fluid/tests/*:F401

.gitignore

@@ -68,6 +68,7 @@ paddle/infrt/tests/dialect/Output
 paddle/infrt/tests/lit.cfg.py
 paddle/infrt/kernel/phi/infershaped/infershaped_kernel_launchers.cc
 paddle/fluid/pybind/eager_op_function.cc
+tools/nvcc_lazy
 
 # these files (directories) are generated before build system generation
 paddle/fluid/operators/generated_op.cc

.pre-commit-config.yaml

@@ -1,8 +1,10 @@
+# Exclude all third-party libraries globally
 exclude: |
     (?x)^(
         patches/.+|
         paddle/fluid/framework/fleet/heter_ps/cudf/.+|
-        paddle/fluid/distributed/ps/thirdparty/round_robin.h
+        paddle/fluid/distributed/ps/thirdparty/round_robin.h|
+        python/paddle/utils/gast/.+
     )$
 repos:
 -   repo: https://github.com/Lucas-C/pre-commit-hooks.git
@@ -12,16 +14,22 @@ repos:
     -   id: remove-tabs
         files: \.(c|cc|cxx|cpp|cu|h|hpp|hxx|xpu|kps)$
         args: [--whitespaces-count, '2']
+    -   id: remove-tabs
+        files: (.*\.(py|bzl)|BUILD|.*\.BUILD|WORKSPACE)$
+        args: [--whitespaces-count, '4']
+        # Exclude the fluid directory but keep the fluid/tests directory.
+        # And exclude some unit test files that require tabs.
+        exclude: |
+            (?x)^(
+                python/paddle/fluid/(?!tests).+|
+                python/paddle/fluid/tests/unittests/collective/fleet/test_hdfs1.py|
+                python/paddle/fluid/tests/unittests/dygraph_to_static/test_error.py
+            )$
 -   repo: https://github.com/google/yapf
     rev: v0.32.0
     hooks:
     -   id: yapf
         files: (.*\.(py|bzl)|BUILD|.*\.BUILD|WORKSPACE)$
-        exclude: |
-            (?x)^(
-                python/paddle/fluid/tests/unittests/dygraph_to_static/test_error.py|
-                python/paddle/fluid/tests/unittests/dygraph_to_static/test_origin_info.py
-            )$
 -   repo: https://github.com/PyCQA/flake8
     rev: 4.0.1
     hooks:

AUTHORS.md

@@ -12,6 +12,7 @@ This is an incomplete list of authors of [Paddle](https://github.com/PaddlePaddl
 | beckett1124 | Bin Qi |
 | ChengduoZH | Cheng-Duo Zhao|
 | chengxiaohua1105 | Xiao-Hua Cheng |
+| chenwhql | Wei-Hang Chen |
 | cxwangyi, yiwangbaidu, wangkuiyi | Yi Wang |
 | cxysteven | Xing-Yi Cheng |
 | ddokupil | Dariusz Dokupil |
@@ -91,3 +92,25 @@ This is an incomplete list of authors of [Paddle](https://github.com/PaddlePaddl
 | jzhang533 | Jun Zhang |
 | Ligoml | Meng-Liu Li |
 | jeff41404 | Xiang Gao |
+| zh794390558 | Hui Zhang |
+| limin2021 | Min Li |
+| zhouwei25 | Wei Zhou |
+| littletomatodonkey | Ruo-Yu Guo |
+| zhupengyang | Zhu Pengyang |
+| DesmonDay | Siming Dai |
+| thisjiang | jiangcheng |
+| yghstill | Guanghua Yu |
+| CtfGo | Tefeng Chen |
+| ZHUI | Hui Zhong|
+| LemonNoel | Huijuan Wang |
+| wawltor | Zeyang Fang |
+| FrostML | Zheng-Xi Liu |
+| jiangjiajun | jiangjiajun |
+| dingjiaweiww | dingjiawei |
+| gglin001 | Allen Guo (Graphcore) |
+| yaozhixin | Zhixin Yao (Graphcore) |
+| XBWGC | Xiaobing Wang (Graphcore) |
+| jianghaicheng | Haicheng Jiang (Graphcore) |
+| czr-gc | Zhaorui Chen (Graphcore) |
+| zhao-han | Han Zhao (Graphcore) |
+| yiakwy, yiakwy-xpu-ml-framework-team | Yi Wang (Graphcore) |

CMakeLists.txt

@@ -323,7 +323,7 @@ unset(WITH_RECORD_BUILDTIME CACHE)
 
 # PY_VERSION
 if(NOT PY_VERSION)
-  set(PY_VERSION 3.6)
+  set(PY_VERSION 3.7)
 endif()
 set(PYBIND11_PYTHON_VERSION ${PY_VERSION})
 

cmake/experiments/cuda_module_loading_lazy.cmake

@@ -38,6 +38,7 @@ if(LINUX)
   set(CUDA_USE_STATIC_CUDA_RUNTIME
       OFF
       CACHE BOOL "" FORCE)
+  set(CMAKE_CUDA_FLAGS "--cudart shared")
   enable_language(CUDA)
   execute_process(
     COMMAND "rm" "-rf" "${CMAKE_SOURCE_DIR}/tools/nvcc_lazy"

cmake/external/python.cmake

@@ -14,6 +14,9 @@
 
 include(python_module)
 
+check_py_version(${PY_VERSION})
+
+# Find Python with mnimum PY_VERSION specified or will raise error!
 find_package(PythonInterp ${PY_VERSION} REQUIRED)
 find_package(PythonLibs ${PY_VERSION} REQUIRED)
 
@@ -79,5 +82,6 @@ if(PYTHONINTERP_FOUND)
         "please use pip to upgrade protobuf. pip install -U protobuf")
   endif()
 endif(PYTHONINTERP_FOUND)
+
 include_directories(${PYTHON_INCLUDE_DIR})
 include_directories(${PYTHON_NUMPY_INCLUDE_DIR})

cmake/external/xpu.cmake

@@ -10,7 +10,7 @@ set(XPU_RT_LIB_NAME "libxpurt.so")
 if(NOT DEFINED XPU_BASE_URL)
   set(XPU_BASE_URL_WITHOUT_DATE
       "https://baidu-kunlun-product.su.bcebos.com/KL-SDK/klsdk-dev")
-  set(XPU_BASE_URL "${XPU_BASE_URL_WITHOUT_DATE}/20220919")
+  set(XPU_BASE_URL "${XPU_BASE_URL_WITHOUT_DATE}/20220928")
 else()
   set(XPU_BASE_URL "${XPU_BASE_URL}")
 endif()
@@ -19,7 +19,7 @@ endif()
 if(NOT DEFINED XPU_XDNN_BASE_URL)
   set(XPU_XDNN_BASE_URL_WITHOUT_DATE
       "https://klx-sdk-release-public.su.bcebos.com/xdnn/dev")
-  set(XPU_XDNN_BASE_URL "${XPU_XDNN_BASE_URL_WITHOUT_DATE}/20220919")
+  set(XPU_XDNN_BASE_URL "${XPU_XDNN_BASE_URL_WITHOUT_DATE}/20220928")
 else()
   set(XPU_XDNN_BASE_URL "${XPU_XDNN_BASE_URL}")
 endif()

cmake/phi.cmake

@@ -78,7 +78,7 @@ function(kernel_declare TARGET_LIST)
     string(
       REGEX
         MATCH
-        "(PD_REGISTER_KERNEL|PD_REGISTER_GENERAL_KERNEL)\\([ \t\r\n]*[a-z0-9_]*,[ \t\r\n\/]*[a-z0-9_]*"
+        "(PD_REGISTER_KERNEL|PD_REGISTER_GENERAL_KERNEL)\\([ \t\r\n]*[a-z0-9_]*,[[ \\\t\r\n\/]*[a-z0-9_]*]?[ \\\t\r\n]*[a-zA-Z]*,[ \\\t\r\n]*[A-Z_]*"
         first_registry
         "${kernel_impl}")
     if(NOT first_registry STREQUAL "")
@@ -89,38 +89,23 @@ function(kernel_declare TARGET_LIST)
           continue()
         endif()
       endif()
-      # parse the first kernel name
-      string(REPLACE "PD_REGISTER_KERNEL(" "" kernel_name "${first_registry}")
-      string(REPLACE "PD_REGISTER_GENERAL_KERNEL(" "" kernel_name
-                     "${kernel_name}")
-      string(REPLACE "," "" kernel_name "${kernel_name}")
-      string(REGEX REPLACE "[ \t\r\n]+" "" kernel_name "${kernel_name}")
-      string(REGEX REPLACE "//cuda_only" "" kernel_name "${kernel_name}")
+      # parse the registerd kernel message
+      string(REPLACE "PD_REGISTER_KERNEL(" "" kernel_msg "${first_registry}")
+      string(REPLACE "PD_REGISTER_GENERAL_KERNEL(" "" kernel_msg
+                     "${kernel_msg}")
+      string(REPLACE "," ";" kernel_msg "${kernel_msg}")
+      string(REGEX REPLACE "[ \\\t\r\n]+" "" kernel_msg "${kernel_msg}")
+      string(REGEX REPLACE "//cuda_only" "" kernel_msg "${kernel_msg}")
+
+      list(GET kernel_msg 0 kernel_name)
+      list(GET kernel_msg 1 kernel_backend)
+      list(GET kernel_msg 2 kernel_layout)
+
       # append kernel declare into declarations.h
-      # TODO(chenweihang): default declare ALL_LAYOUT for each kernel
-      if(${kernel_path} MATCHES "./cpu\/")
-        file(APPEND ${kernel_declare_file}
-             "PD_DECLARE_KERNEL(${kernel_name}, CPU, ALL_LAYOUT);\n")
-      elseif(${kernel_path} MATCHES "./gpu\/")
-        file(APPEND ${kernel_declare_file}
-             "PD_DECLARE_KERNEL(${kernel_name}, GPU, ALL_LAYOUT);\n")
-      elseif(${kernel_path} MATCHES "./xpu\/")
-        file(APPEND ${kernel_declare_file}
-             "PD_DECLARE_KERNEL(${kernel_name}, XPU, ALL_LAYOUT);\n")
-      elseif(${kernel_path} MATCHES "./gpudnn\/")
-        file(APPEND ${kernel_declare_file}
-             "PD_DECLARE_KERNEL(${kernel_name}, GPUDNN, ALL_LAYOUT);\n")
-      elseif(${kernel_path} MATCHES "./kps\/")
-        file(APPEND ${kernel_declare_file}
-             "PD_DECLARE_KERNEL(${kernel_name}, KPS, ALL_LAYOUT);\n")
-      elseif(${kernel_path} MATCHES "./onednn\/")
-        file(APPEND ${kernel_declare_file}
-             "PD_DECLARE_KERNEL(${kernel_name}, OneDNN, ALL_LAYOUT);\n")
-      else()
-        # deal with device independent kernel, now we use CPU temporaary
-        file(APPEND ${kernel_declare_file}
-             "PD_DECLARE_KERNEL(${kernel_name}, CPU, ALL_LAYOUT);\n")
-      endif()
+      file(
+        APPEND ${kernel_declare_file}
+        "PD_DECLARE_KERNEL(${kernel_name}, ${kernel_backend}, ${kernel_layout});\n"
+      )
     endif()
   endforeach()
 endfunction()

cmake/python_module.cmake

@@ -48,3 +48,20 @@ function(find_python_module module)
       ${PY_${module_upper}_VERSION}
       PARENT_SCOPE)
 endfunction()
+
+function(check_py_version py_version)
+  string(REPLACE "." ";" version_list ${py_version})
+  list(LENGTH version_list version_list_len)
+  if(version_list_len LESS 2)
+    message(FATAL_ERROR "Please input Python version, eg:3.7 or 3.8 and so on")
+  endif()
+
+  list(GET version_list 0 version_major)
+  list(GET version_list 1 version_minor)
+
+  if((version_major GREATER_EQUAL 3) AND (version_minor GREATER_EQUAL 7))
+
+  else()
+    message(FATAL_ERROR "Paddle only support Python version >=3.7 now!")
+  endif()
+endfunction()