I was working with K80 architecture Tesla, and my code (For educational simulations) was written in Cuda 8.
Now I want to upgrade my Gpu card to RTX 2080 with the architecture Turing and it’s compatible with Cuda 10.
My question is, can I run my old code (which is written in cuda 8 and works with Tesla architecture), in this new GPU card RTX 2080.
In theory it should, though not having tested this particular case myself, I can’t guarantee it. Would you be able to recompile the code to target the newer CUDA version and device?
Thank you for your answer, yeah in theory it works, and I try it today fortunately it works, there is no problem.