GPU code generation environment test fails

Hello,

i'm running GPU code generation environment test, and it fails:

>> coder.checkGpuInstall('full')
Compatible GPU           : PASSED 
CUDA Environment         : PASSED 
	Runtime   : PASSED 
	cuFFT     : PASSED 
	cuSOLVER  : PASSED 
	cuBLAS    : PASSED 
cuDNN Environment        : PASSED 
TensorRT Environment     : PASSED 
Profiling Environment    : PASSED 
Basic Code Generation    : PASSED 
Basic Code Execution     : PASSED 
Deep Learning (TensorRT) Code Generation: PASSED 
TensorRT INT8 Compute Capability Check: PASSED 
Deep Learning (TensorRT) Code Execution: FAILED (Unable to find TensorRT libraries in 'PATH' environment variable for mex target. Add TensorRT library path to 'PATH' before launching MATLAB.)

I added PATH C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\TensorRT-5.1.5.0\lib to environment variables, but still the same result. I have checked all the prerequistes according to https://fr.mathworks.com/help/gpucoder/gs/setting-up-the-toolchain.html

Please help.

Answer 1

[Robert van der Willigen]

NVIDIA_TENSORRT parameter is probably not set see below:

% Selecting Gpu 
gpuDevice(1);

% CUDA Installation Path 
setenv('CUDA_PATH', ['C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1' pathsep getenv('CUDA_PATH')]);

% cuDNN Installation Path 
setenv('NVIDIA_CUDNN','C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\cudnn-10.1-windows10-x64-v8.0.4.30');

% TensorRT Installation Path 
setenv('NVIDIA_TENSORRT','C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\TensorRT-6.0.1.5.Windows10.x86_64.cuda-10.1.cudnn7.6');

% NVTX Library Path 
setenv('NVTOOLSEXT_PATH','C:\Program Files\NVIDIA Corporation\NvToolsExt\');

Answer 2

[Robert van der Willigen]

also have a look at all the parameters that need to be set:

https://nl.mathworks.com/help/gpucoder/gs/setting-up-the-toolchain.html