Hello,
Trying to understand what performance to expect from Frame Generation and at what point I will hit a bottleneck preventing me from reaching 2X rates. The Super Resolution SDK has a really cool table to give ballpark ideas of the execution time for various resolutions/GPUs:
Any chance there are some rough performance indicators for Frame Generation? For example, for a given resolution (say 4K), how much time does it take to interpolate the frame? Trying to figure out based on application frame times, how to achieve 2X frame rates and what the limiting factors would be, in particular if I am not using Super Resolution in conjunction with Frame Generation.
I’ve tried making some rough measurements myself around the feature evaluation time, but the duration seems a bit high, so I suspect there might be something wrong with what I am doing. The ballpark figures of what to expect would help me understand whether or not I need to dig deeper.
Thanks!