I have been trying to install Cuda for a quite a few hours now. I need it for GPU support in Deep Learning Studio (Deep Cognition).
Deep Learning Studio says “GPU not supported”.
What I had installed was: Visual Studio 2019, Cuda 9.0, current Geforce Experience (Version 441.66) and Deep Learning Studio on Windows 10
When reinstalling DLS and Cuda, the Cuda installer said “this graphics driver could not find compatible graphics hardware”, so i searched and found that I need to unclick “install graphics driver”, because the driver that comes with Cuda 9 is too old for a RTX 2060 Super. So I did that and reinstalled everything.
I wanted to try Samples from Cuda now, so I went into the folder and found that there were only samples for VS 2017 and older, so i uninstalled VS 2019 and reinstalled everything again with VS 2017.
When I open the samples now (deviceQuery_vs2017.sln in my case) and try to run it, it shows tons of errors with the codes: E1696, E0282, E0020, and many more.
It seems that it can’t open something.
Maybe I have to copy something in PATH or something like that. I don’t know about that stuff.
Also when installing Cuda, at the end it says:
“graphics driver: Version: 385.54, Status: Not selected”
My Geforce Experience has the driver Version 441.66.
I also read the official installation guide for CUDA by Nvidia, but it doesn’t help.
Could you maybe give me foolproof instructions on how you would install Cuda 9.0 for an RTX 2060 Super on a fresh install of Windows 10 (so including installing Visual Studio and other programs which you need)?
I am really having trouble installing GPU support for DLS. I would be very thankful for help.
And I have a question about the whole installation process: Why is this so hard and not a simple installation like programs that run on the CPU (like Deep Learning Studio with CPU support)?