Git cuda samples wont compile. nvcc reports "compute_89 wrong architecture"

I have a fresh install of Ubuntu 22.04 . My Nvidia device is a Quadro P3000 Mobile. I installed the 515 driver and 11.7 cuda toolkit. Followed these instructions…

installation seemed to go smooth. rebooted. nvcc -V reports cuda 11.7 as expected. However nvidia-smi reports cuda 12.1 and a driver version of 530.30.2. Down loaded some cuda samples from Git. when I compile the samples I get the error compute_89 wrong architecture". Not sure how this is happening but this tool chain seems to be the wrong one. Can any one advise on what the proper chain is and where to get it?

The cuda version reported in nvidia-smi is the cuda version that both nvidia-smi and the driver were compiled with - nvidia-smi being part of the driver install.

Your GPU is SM 6.1. What happens when you compile your samples:

make SMS=“61”?

Edited: To correct quotation placement.

That worked! Thank you very much! For my education, where is it documented that my gpu is an SM 6.1? Thanks again!

In this instance, I used this reference, under “Cuda:”.