Merge remote-tracking branch 'origin/master' into cnet

This commit is contained in:
Davide Sapienza
2021-04-27 14:43:39 +02:00
43 changed files with 1835 additions and 148 deletions
+12 -4
View File
@@ -2,7 +2,14 @@ cmake_minimum_required(VERSION 3.5)
project (tkDNN)
set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} ${CMAKE_CURRENT_SOURCE_DIR}/cmake)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11 -fPIC -Wno-deprecated-declarations -Wno-unused-variable")
if(UNIX)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11 -fPIC -Wno-deprecated-declarations -Wno-unused-variable ")
endif()
if(WIN32)
set(CMAKE_CXX_STANDARD 11)
set(CMAKE_CXX_FLAGS "/O2 /FS /EHsc")
set(CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS ON)
endif(WIN32)
include_directories(${CMAKE_CURRENT_SOURCE_DIR}/include/tkDNN)
# project specific flags
@@ -18,7 +25,7 @@ add_definitions(-DTKDNN_PATH="${CMAKE_CURRENT_SOURCE_DIR}")
find_package(CUDA 9.0 REQUIRED)
SET(CUDA_SEPARABLE_COMPILATION ON)
#set(CUDA_NVCC_FLAGS "${CUDA_NVCC_FLAGS} -arch=sm_30 --compiler-options '-fPIC'")
set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} --maxrregcount=32)
set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} --maxrregcount=32 -arch=sm_61 )
find_package(CUDNN REQUIRED)
include_directories(${CUDNN_INCLUDE_DIR})
@@ -28,6 +35,7 @@ include_directories(${CUDNN_INCLUDE_DIR})
file(GLOB tkdnn_CUSRC "src/kernels/*.cu" "src/sorting.cu")
cuda_include_directories(${CMAKE_CURRENT_SOURCE_DIR}/include ${CUDA_INCLUDE_DIRS} ${CUDNN_INCLUDE_DIRS})
cuda_add_library(kernels SHARED ${tkdnn_CUSRC})
target_link_libraries(kernels ${CUDA_CUBLAS_LIBRARIES})
#-------------------------------------------------------------------------------
@@ -40,7 +48,7 @@ find_package(OpenCV REQUIRED)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -DOPENCV")
# gives problems in cross-compiling, probably malformed cmake config
#find_package(yaml-cpp REQUIRED)
find_package(yaml-cpp REQUIRED)
#-------------------------------------------------------------------------------
# Build Libraries
@@ -48,7 +56,7 @@ set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -DOPENCV")
file(GLOB tkdnn_SRC "src/*.cpp")
set(tkdnn_LIBS kernels ${CUDA_LIBRARIES} ${CUDA_CUBLAS_LIBRARIES} ${CUDNN_LIBRARIES} ${OpenCV_LIBS} yaml-cpp)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}")
include_directories(${CMAKE_CURRENT_SOURCE_DIR}/include ${CUDA_INCLUDE_DIRS} ${OPENCV_INCLUDE_DIRS} ${NVINFER_INCLUDES})
add_library(tkDNN SHARED ${tkdnn_SRC})
target_link_libraries(tkDNN ${tkdnn_LIBS})