NO CMAKE_CUDA_COMPILER could be found when building CUDA app on x86

[Nir-Az]

Hi,

I am trying to build an application that use CUDA on x86 PC (using a docker and qemu). I followed your Wiki steps

When I run the CMake configure command I get this error NO CMAKE_CUDA_COMPILER could be found

I c an build it on ARM without CUDA usage, but if I turn on the CUDA usage on my applicatiob CMAKE I get this error The image I use is the one from the Wiki : nvcr.io/nvidia/l4t-base:r32.3.1

The step I take are:

• Install docker

• Install qemu

• Install nvidia docker2

• restart docker service

• run qemu docker

• run Cuda image docker

• verify uname -m = aarch64

• clone application

• install necessary depends

• run CMake

I will add that when looking for cuda on the docker file system I found a symbolic link of cuda to cuda-10.0 under /usr/local but the cuda-10.0 folder is empty. I would appreciate some help regarding the missing CUDA compiler..

** UPDATE**: When running the example from the Wiki I can see that I should expect flis under the Cuda folder and they do not exist. root@7 7616378b275:/# cp -r /usr/local/cuda/samples /tmp /bin/cp: cannot stat '/usr/local/cuda/samples': No such file or folder

Thanks Nir