GPU Coder requirements for Simulink

I'm trying to run GPU coder on Simulink. The Prerequisites says "CUDA-enabled NVIDIA® GPU with compute capability 3.0 or higher.". However, the GPU coder itself requires "NVIDIA® GPU enabled for CUDA with compute capability 3.2 or higher ". I'm using this:

                      Name: 'Quadro K4100M'
                     Index: 1
         ComputeCapability: '3.0'
            SupportsDouble: 1
             DriverVersion: 10.1000
            ToolkitVersion: 10
        MaxThreadsPerBlock: 1024
          MaxShmemPerBlock: 49152
        MaxThreadBlockSize: [1024 1024 64]
               MaxGridSize: [2.1475e+09 65535 65535]
                 SIMDWidth: 32
               TotalMemory: 4.2950e+09
           AvailableMemory: 3.6442e+09
       MultiprocessorCount: 6
              ClockRateKHz: 705500
               ComputeMode: 'Default'
      GPUOverlapsTransfers: 1
    KernelExecutionTimeout: 0
          CanMapHostMemory: 1
           DeviceSupported: 1
            DeviceSelected: 1

And when I do GPU test I get:

Compatible GPU           : FAILED (The compute capability '3.0' of the selected GPU '0' is not supported by GPU Coder. Execution of the generated GPU MEX will not be available.)
CUDA Environment         : PASSED 
	Runtime   : PASSED 
	cuFFT     : PASSED 
	cuSOLVER  : PASSED 
	cuBLAS    : PASSED 
cuDNN Environment        : PASSED 
Basic Code Generation    : FAILED (Test GPU code generation failed with the error 'emlc:compilationError'. View report for further information: View report)

I'm using matlab 2019b with Microsoft Visual C++ 2015 as compiler.

Anyone can help?

Answer 1

[Matteo Meli]

Same problem here except I've used Visual C++ 2019 as compiler. I also tried changing the compiler via mex -setup. No success.

Name: 'GeForce GTX 1080'
                     Index: 1
         ComputeCapability: '6.1'
            SupportsDouble: 1
             DriverVersion: 10.2000
            ToolkitVersion: 10.1000
        MaxThreadsPerBlock: 1024
          MaxShmemPerBlock: 49152
        MaxThreadBlockSize: [1024 1024 64]
               MaxGridSize: [2.1475e+09 65535 65535]
                 SIMDWidth: 32
               TotalMemory: 8.5899e+09
           AvailableMemory: 6.9973e+09
       MultiprocessorCount: 20
              ClockRateKHz: 1822500
               ComputeMode: 'Default'
      GPUOverlapsTransfers: 1
    KernelExecutionTimeout: 1
          CanMapHostMemory: 1
           DeviceSupported: 1
            DeviceSelected: 1    

coder.checkGpuInstall('full') gives me this :

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    : FAILED (Test GPU code generation failed with the error 'emlc:compilationError'. View report for further information: View report)
Deep Learning (TensorRT) Code Generation: FAILED (Test GPU code generation failed with the error 'emlc:compilationError'. View report for further information: View report)
TensorRT INT8 Compute Capability Check: PASSED 

ans = 

  struct with fields:

                 gpu: 1
                cuda: 1
               cudnn: 1
            tensorrt: 1
        basiccodegen: 0
       basiccodeexec: 0
         deepcodegen: 0
        deepcodeexec: 0
    tensorrtdatatype: 1
           profiling: 1

Answer 2

[Matteo Meli]

Hello,

I've found a rather odd workaround to get to 'Passed' for all entries.

Here was what I did:

1) Install Microsoft Visual C++ 2017 as compiler

2) Create a new Windows User Account

3) Create a new MATLAB Licence for that new user

4) Load the new license into MATLAB

5) Check if coder.checkGpuInstall('full') gives you a 'Passed' for all entries.

I would like to hear your feedback if this solution also works for you!

Best regards,

Matteo Meli