MATLAB not detecting Nvidia RTX 1050Ti on WIndows 11

Hello friends,
I am working with MATLAB R2020b on Window 11, Nvidia GeForce RTX 1050 Ti. I have installed the latest Nvidia driver, as well as CUDA 10.2. I get the error below after running
gpuDevice
Error using gpuDevice (line 26)
Unable to find a supported GPU device. For more information on GPU support, see GPU Support by Release.
Can someone kindly help me resolve this problem? Thanks

[Joss Knight]

Please run nvidia-smi in a command window or power shell and paste the output here.

[Richard Fiifi Annan]

Thank you @Joss Knight for the response. Below is the output:

+-----------------------------------------------------------------------------+

| NVIDIA-SMI 512.96 Driver Version: 512.96 CUDA Version: 11.6 |

|-------------------------------+----------------------+----------------------+

| GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. ECC |

| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |

| | | MIG M. |

|===============================+======================+======================|

| 0 NVIDIA GeForce ... WDDM | 00000000:01:00.0 Off | N/A |

| N/A 48C P8 6W / N/A | 719MiB / 4096MiB | 0% Default |

| | | N/A |

+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+

| Processes: |

| GPU GI CI PID Type Process name GPU Memory |

| ID ID Usage |

|=============================================================================|

| 0 N/A N/A 9276 C+G ...020b\bin\win64\MATLAB.exe N/A |

+-----------------------------------------------------------------------------+

[Joss Knight]

I'm not seeing an obvious reason here. Is this a laptop? If not you may have to open an issue with tech support.

[Richard Fiifi Annan]

Yes @Joss Knight, it is a laptop. Do I still have to open an issue?

[Joss Knight]

Ah, well then this could be the hardware acceleration settings. Some laptops require you to explicitly whitelist applications for external GPUs because of power saving.

Under Graphics settings-> Graphics performance preference, add MATLAB to the list and restart.

[Richard Fiifi Annan]

I have done it, but still nothing has changed. Same error message

[Joss Knight]

Screenshot of the settings window?

[Joss Knight]

You might try the same settings in the NVIDIA Control Panel (right-click on the desktop -> NVIDIA Control Panel), under Manage 3D Settings:

[Richard Fiifi Annan]

Hello @Joss Knight, I am grateful for the responses; however, the problem is still not resolved.I have uploaded screenshot of the settings I have made.

[Richard Fiifi Annan]

This the screenshot of the graphic settings

[Richard Fiifi Annan]

Hello @Joss Knight, I think @Walter Roberson has allayed my worries. Thank very much for the responses

Answer 1

[Walter Roberson]

Your post says Nvidia GeForce RTX 1050 Ti, but that black image says RTX 3050 Ti. That could make a notable difference in your R2020b release. IIRC Ampere was not supported in R2020b.

https://www.mathworks.com/help/releases/R2021b/parallel-computing/gpu-support-by-release.html

You have to enable support for Ampere in R2020b.

[Richard Fiifi Annan]

You are right @Walter Roberson. I admit I made a typo in the post title. I was previously using 1050Ti laptop, so I guess that how the confusion came in. I honestly do not know how to enable Ampere for R2020b. Can you please guide me through it? Thanks for your help.

[Walter Roberson]

https://www.mathworks.com/help/releases/R2021b/parallel-computing/gpu-support-by-release.html#mw_aebb8064-40ea-447f-ac9a-ec875be27224

[Richard Fiifi Annan]

Oh sorry, I found the answer in the web link you provided. Sincerely grateful

[Richard Fiifi Annan]

Hello friends @Joss Knight and @Walter Roberson, the problem is solved. I had to enable forward compatibilty using

parallel.gpu.enableCUDAForwardCompatibility(1)

[Joss Knight]

You must have been hitting a bug because MATLAB is supposed to display an error that directs you to the forward compatibility feature.

I hope after all this you get some benefit, a laptop GPU using forward compatibility - you might not see much better performance than the CPU.

[Richard Fiifi Annan]

Hi @Joss Knight, what you said is painfully true. I ran some deep learning regression after forward compatibility; and the RMSE was virtually constant after about 10 epochs. I have to resort to an older PC. Once again, thanks very much.

[Joss Knight]

Hiya. I was only expecting that your GPU's performance might not be that great - but it wouldn't make any difference to whether or not a model will work, only how long it takes to train. If you're not seeing any improvement when training your model, maybe you need to change the model or the training parameters.