tkDNN can now deserialize tensorrt-8 engine (both through test_* and trtexec)
but demo has issues in yolo::computeDetections
This commit is contained in:
+6
-4
@@ -31,7 +31,7 @@ endif()
|
||||
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 -G -g)
|
||||
set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS} --maxrregcount=32)
|
||||
|
||||
|
||||
find_package(CUDNN REQUIRED)
|
||||
@@ -39,8 +39,10 @@ include_directories(${CUDNN_INCLUDE_DIR})
|
||||
|
||||
|
||||
# compile
|
||||
file(GLOB tkdnn_CUSRC "src/kernels/*.cu" "src/sorting.cu" )
|
||||
file(GLOB tkdnn_CUSRC "src/kernels/*.cu" "src/sorting.cu" "src/pluginsRT/*.cpp")
|
||||
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_LIBRAY} ${CUDA_LIBRARIES} ${CUDNN_LIBRARIES})
|
||||
|
||||
|
||||
|
||||
@@ -64,11 +66,11 @@ find_package(yaml-cpp REQUIRED)
|
||||
# Build Libraries
|
||||
#-------------------------------------------------------------------------------
|
||||
file(GLOB tkdnn_SRC "src/*.cpp")
|
||||
set(tkdnn_LIBS ${CUDA_LIBRARIES} ${CUDA_CUBLAS_LIBRARIES} ${CUDNN_LIBRARIES} ${OpenCV_LIBS} yaml-cpp)
|
||||
set(tkdnn_LIBS kernels ${CUDA_LIBRARIES} ${CUDA_CUBLAS_LIBRARIES} ${CUDNN_LIBRARIES} ${OpenCV_LIBS} yaml-cpp)
|
||||
|
||||
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}")
|
||||
include_directories(${CMAKE_CURRENT_SOURCE_DIR}/include ${CUDA_INCLUDE_DIRS} ${OPENCV_INCLUDE_DIRS} ${NVINFER_INCLUDES})
|
||||
cuda_add_library(tkDNN SHARED ${tkdnn_SRC} ${tkdnn_CUSRC})
|
||||
add_library(tkDNN SHARED ${tkdnn_SRC})
|
||||
target_link_libraries(tkDNN ${tkdnn_LIBS} ${CUDA_CUBLAS_LIBRAY})
|
||||
|
||||
#static
|
||||
|
||||
Reference in New Issue
Block a user