I have tried to complied darknet on my Jetson tx2 on Auvidea J121 carrier board
I complied darknet without GPU ,CUDA ,and cuDNN. There's no error.
Once I tried to compile with them, I got this error
./src/network_kernel.cu(706): error :identifier "cudaStreamCaptureModelGlobal" is undifined
./src/network_kernel.cu(706): error : too many arguments in function call
2 errors detected in the compilaiton of "/tmp/tmpxft_00005cdb_00000000-9_network_kernels.compute_61.cpp.ii".
Makefile:185: recipe for target 'obj/network_kernels.o' failed
make: ***[obj/network_kernels.o] Error 1
CUDA and cuDNN were installed and works properly on its test code.
I have another Tx2 module on its develop carrier. It can compile darknet with no problem and can run it on 30+ fps.
I have tried to complied darknet on my Jetson tx2 on Auvidea J121 carrier board I complied darknet without GPU ,CUDA ,and cuDNN. There's no error. Once I tried to compile with them, I got this error
./src/network_kernel.cu(706): error :identifier "cudaStreamCaptureModelGlobal" is undifined ./src/network_kernel.cu(706): error : too many arguments in function call
2 errors detected in the compilaiton of "/tmp/tmpxft_00005cdb_00000000-9_network_kernels.compute_61.cpp.ii". Makefile:185: recipe for target 'obj/network_kernels.o' failed make: ***[obj/network_kernels.o] Error 1
CUDA and cuDNN were installed and works properly on its test code. I have another Tx2 module on its develop carrier. It can compile darknet with no problem and can run it on 30+ fps.