Well thats the fact, i have been able to install several version of CUDA on Linux and a way to change between them however MATLAB does not perform the change i required the 9.1 version to Cross-Compiling developer.
Already solve this, running everything from a source, so basically my question is do i need to pass all the tests from coder.checkGpuInstall(‘full’) because at the time the setup of cuDNN is a headache and the setup of TensporRT well is the same ,can i skip it if iḿ not planning to use TensorRT and cuDNN otherwise i have several CUDA installations how to fix the calling to the specified CUDNN setup? the thing is that iḿ getting this error:
cuDNN Environment : FAILED (Unable to find cuDNN header files in directory ‘/usr/local/cuda-9.1/targets/x86_64-linux/lib/libcudnn.so.7:/usr/local/cuda/:/usr/local/cuda-9.1/targets/x86_64-linux/lib:/usr/local/cuda-9.1/targets/x86_64-linux/include:/usr/local/cuda-9.1/include:/usr/local/cuda-9.1/include:/usr/local/cuda-9.1/lib64:/usr/local/cuda/lib64:/usr/local/cuda-9.1/lib64:/usr/local/cuda/lib:/include’. Check that the cuDNN headers are installed with the specified cuDNN SDK.)
And i use the specified CUDA version for the .deb files, so i dont know whats wrong
Just as remember my goal is crosscompile on Jetson TX2