Merge branch 'master' of https://github.com/apache/incubator-mxnet

* 'master' of https://github.com/apache/incubator-mxnet: (192 commits)
  * impl - FFI for np einsum (apache#17869)
  [Numpy] FFI for diag/diagonal/diag_indices_from (apache#17789)
  [Numpy] Kron operator (apache#17323)
  cmake: Set DMLC_LOG_FATAL_THROW only for building mxnet and not for tvm (apache#17878)
  Add simplified HybridBlock.forward without F (apache#17530)
  Use FP32 copy of weights for norm (multitensor LAMB optimizer) (apache#17700)
  Use multi-tensor sumSQ in clip_global_norm (apache#17652)
  [Numpy] Add op fmax, fmin, fmod (apache#17567)
  Adding sparse support to MXTensor for custom operators (apache#17569)
  Update 3rdparty/mkldnn to v1.2.2 (apache#17313)
  Dynamic subgraph compile support (apache#17623)
  Refactor cpp-package CMakeLists.txt & add missing inference/imagenet_inference (apache#17835)
  staticbuild: Fix potential user-assisted execution of arbitrary code  (apache#17860)
  * FFI for np.argmax and np.argmin (apache#17843)
  ffi for roll/rot90 (apache#17861)
  Skip test_multi_worker_dataloader_release_pool on OS X (apache#17797)
  add ffi for full_like, binary (apache#17811)
  HybridBlock.export() to return created filenames (apache#17758)
  Fix SoftReLU fused operator numerical stability (apache#17849)
  CI: Test clang10 cpu & gpu builds with -WError (apache#17830)
  ...

.github/workflows/os_x_staticbuild.yml

@@ -0,0 +1,26 @@
+name: continuous build
+
+on: [push, pull_request]
+
+jobs:
+  macosx-x86_64:
+    runs-on: macos-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v2
+      - name: Install Dependencies
+        run: |
+          brew install nasm automake ninja libtool cmake pkgconfig protobuf
+      - name: Build project
+        run: |
+          git --version
+          clang --version
+          CMAKE_STATICBUILD=1 ./tools/staticbuild/build.sh cpu
+      - name: Setup Python
+        run: |
+          python3 -m pip install --user nose nose-timer nose-exclude numpy scipy
+          python3 -m pip install --user -e python
+      - name: Test project
+        run: |
+          python3 -m nose --with-timer --verbose tests/python/unittest/ --exclude-test=test_extensions.test_subgraph --exclude-test=test_extensions.test_custom_op --exclude-test=test_gluon_data.test_recordimage_dataset_with_data_loader_multiworker
+          

.gitignore

@@ -36,6 +36,7 @@ docs/web-data
 
 #dmlc
 config.mk
+config.cmake
 
 *.pyc
 .Rhistory

.gitmodules

@@ -16,7 +16,6 @@
 [submodule "3rdparty/mkldnn"]
 	path = 3rdparty/mkldnn
 	url = https://github.com/intel/mkl-dnn.git
-	branch = master
 [submodule "3rdparty/tvm"]
 	path = 3rdparty/tvm
 	url = https://github.com/apache/incubator-tvm.git

3rdparty/mshadow/CMakeLists.txt

@@ -1,6 +1,87 @@
-cmake_minimum_required(VERSION 2.8.7)
-
+cmake_minimum_required(VERSION 3.13)
 project(mshadow C CXX)
 
+include(CMakeDependentOption)
+option(USE_CUDA "Build with CUDA support" ON)
+option(USE_CUDNN ON)
+cmake_dependent_option(USE_SSE "Build with x86 SSE instruction support" ON "NOT ARM" OFF)
+option(USE_F16C "Build with x86 F16C instruction support" ON) # autodetects support if ON
+option(USE_INT64_TENSOR_SIZE "Use int64_t to represent the total number of elements in a tensor" OFF)
+option(MSHADOW_IN_CXX11 ON)
+
+add_library(mshadow INTERFACE)
+file(GLOB_RECURSE MSHADOWSOURCE "mshadow/*.h")
+target_include_directories(mshadow INTERFACE "${CMAKE_CURRENT_SOURCE_DIR}")
+target_sources(mshadow INTERFACE ${MSHADOWSOURCE})
+if(UNIX)
+  target_compile_options(mshadow INTERFACE
+    "$<$<AND:$<COMPILE_LANGUAGE:CXX>,$<CXX_COMPILER_ID:Clang>>:-Wno-braced-scalar-init>"
+    "$<$<AND:$<COMPILE_LANGUAGE:CXX>,$<CXX_COMPILER_ID:Clang>>:-Wno-pass-failed>"
+    # TODO Replace Wno-unused-lambda-capture with [[maybe_unused]] annotation once requiring C++17
+    "$<$<AND:$<COMPILE_LANGUAGE:CXX>,$<CXX_COMPILER_ID:Clang>>:-Wno-unused-lambda-capture>"
+    # TODO Fixing the warning leads to compile error on 4.8; fix once 4.8 support is dropped
+    "$<$<AND:$<COMPILE_LANGUAGE:CXX>,$<CXX_COMPILER_ID:Clang>>:-Wno-undefined-var-template>"
+    "$<$<COMPILE_LANGUAGE:CXX>:-Wno-unused-parameter>"
+    "$<$<COMPILE_LANGUAGE:CXX>:-Wno-unknown-pragmas>"
+    "$<$<COMPILE_LANGUAGE:CXX>:-Wno-unused-local-typedefs>"
+    "$<$<COMPILE_LANGUAGE:CUDA>:--expt-relaxed-constexpr>")
+endif()
+
+if(USE_CUDA)
+  enable_language(CUDA)
+  file(GLOB_RECURSE MSHADOW_CUDASOURCE "mshadow/*.cuh")
+  target_sources(mshadow INTERFACE ${MSHADOW_CUDASOURCE})
+  target_compile_definitions(mshadow INTERFACE MSHADOW_USE_CUDA=1
+    MSHADOW_FORCE_STREAM)
+else()
+  target_compile_definitions(mshadow INTERFACE MSHADOW_USE_CUDA=0)
+endif()
+if(USE_SSE)
+  # For cross compilation, we can't rely on the compiler checks, but mshadow
+  # will add platform specific includes not available in other arches
+  include(CheckCXXCompilerFlag)
+  check_cxx_compiler_flag("-msse3" SUPPORT_MSSE3)
+  check_cxx_compiler_flag("-msse2" SUPPORT_MSSE2)
+  if(SUPPORT_MSSE3)
+    target_compile_definitions(mshadow INTERFACE MSHADOW_USE_SSE)
+    target_compile_options(mshadow INTERFACE $<$<COMPILE_LANGUAGE:CXX>:-msse3>)
+  elseif(SUPPORT_MSSE2)
+    target_compile_definitions(mshadow INTERFACE MSHADOW_USE_SSE)
+    target_compile_options(mshadow INTERFACE $<$<COMPILE_LANGUAGE:CXX>:-msse2>)
+  else()
+    target_compile_definitions(mshadow INTERFACE MSHADOW_USE_SSE=0)
+  endif()
+else()
+  target_compile_definitions(mshadow INTERFACE MSHADOW_USE_SSE=0)
+endif()
+if(USE_CUDNN)
+  target_compile_definitions(mshadow INTERFACE MSHADOW_USE_CUDNN)
+endif()
+if(MSHADOW_IN_CXX11)
+  target_compile_definitions(mshadow INTERFACE MSHADOW_IN_CXX11)
+endif()
+if(USE_F16C)
+  # Determine if hardware supports F16C instruction set
+  message(STATUS "Determining F16C support")
+  include(cmake/AutoDetectF16C.cmake)
+  if(SUPPORT_F16C)
+    target_compile_options(mshadow INTERFACE $<$<COMPILE_LANGUAGE:CXX>:-mf16c>)
+  else()
+    target_compile_definitions(mshadow INTERFACE MSHADOW_USE_F16C=0)
+  endif()
+else()
+  target_compile_definitions(mshadow INTERFACE MSHADOW_USE_F16C=0)
+endif()
+if(USE_INT64_TENSOR_SIZE)
+  message(STATUS "Using 64-bit integer for tensor size")
+  target_compile_definitions(mshadow INTERFACE MSHADOW_INT64_TENSOR_SIZE=1)
+else()
+  target_compile_definitions(mshadow INTERFACE MSHADOW_INT64_TENSOR_SIZE=0)
+endif()
+
 set(mshadow_LINT_DIRS mshadow mshadow-ps)
-add_custom_target(mshadow_lint COMMAND ${CMAKE_COMMAND} -DMSVC=${MSVC} -DPYTHON_EXECUTABLE=${PYTHON_EXECUTABLE} -DLINT_DIRS=${mshadow_LINT_DIRS} -DPROJECT_SOURCE_DIR=${PROJECT_SOURCE_DIR} -DPROJECT_NAME=mshadow -P ${PROJECT_SOURCE_DIR}/../dmlc-core/cmake/lint.cmake)
+find_package(Python3)
+add_custom_target(mshadow_lint COMMAND ${CMAKE_COMMAND} -DMSVC=${MSVC}
+  -DPYTHON_EXECUTABLE=${Python3_EXECUTABLE} -DLINT_DIRS=${mshadow_LINT_DIRS}
+  -DPROJECT_SOURCE_DIR=${PROJECT_SOURCE_DIR} -DPROJECT_NAME=mshadow
+  -P ${PROJECT_SOURCE_DIR}/../dmlc-core/cmake/lint.cmake)

cmake/AutoDetectF16C.cmake → 3rdparty/mshadow/cmake/AutoDetectF16C.cmake

@@ -25,8 +25,11 @@ if(AUTO_DETECT_F16_CMAKE_INCLUDED)
   return()
 endif()
 set(AUTO_DETECT_F16_CMAKE_INCLUDED True)
-
 set(SUPPORT_F16C False)
+if(ANDROID)
+    message("F16C instruction set is not yet supported for Andriod")
+    return()
+endif()
 if(MSVC)
     message("F16C instruction set is not yet supported for MSVC")
     return()