What protocol is used to transfer calculations to the GPU?

When you select "Enable GPU Acceleration" in the CFD Analysis Option Control, it enables the use of GPU computing capabilities. 

To use GPU computation, it should be separate from the display graphics card used for screen output. 

NFX supports Tesla series GPUs (https://en.wikipedia.org/wiki/Nvidia_Tesla) for GPU acceleration. 

It's worth noting that recently, the Tesla series has been rebranded as Ampere. 

NFX supports GPUs with double precision performance of 1000 or more. 

For other GPU models or specifications, they are not supported, and using them may result in errors, so please exercise caution.