I am using nvcc 6.5 under CUDA 11.0 in Visual Studio 2019 with 680 GTX cards. These cards only support API version 1.0.
I have changed the Visual Studio option from “sm_30,compute_30” to:
-gencode=arch=compute_10,code="sm_10,compute_10"
But now receive the build error:
nvcc fatal : Unsupported gpu architecture ‘compute_10’
Do I have the wrong values? Why is this occuring? Note: I was able to compile my code under Visual Studio 2017,but this project must use 2019.