Hi, I apologize in advance for such basic questions. Just got a new Windows 11 laptop with an RTX 4070 mobile GPU and I’d like to explore GPU programming in Python and via C. Trouble is it isn’t clear how to proceed; I really don’t want to mess up the installation. I can run NVIDIA-SMI from a Windows Power Shell and the output shows: Driver Version: 528.66 and Cuda Version: 12.0. Does this mean Nvidia drivers and Cuda are already installed? I’m guessing no because if I type nvcc --version the result is nvcc : The term ‘nvcc’ is not recognized as the name of a cmdlet, function, script file, or operable program. Check the spelling of the name, or if a path was included, verify that the path is correct and try again.
In my searches I’ve found some additional confusing things:
-
I’ve read that the Cuda toolkit can be installed by installing Pytorch within Anaconda, but I have no idea if this will allow me to use the GPU with C (for example in Codeblocks or some other IDE).
-
Looking at the CUDA Installation Guide for Microsoft Windows, the second item under System Requirements is “A supported version of Linux with a gcc compiler and toolchain”. This seems strange because I’m using Windows, not Linux.
-
If I go to the Pytorch site, the installation instructions are for Cuda 11.7 or 11.8, but not 12.0 (which is shown from the Nvidia-smi command)
Can someone suggest what I do from here?