Framerate Isn’t Good Enough: Latency Pipeline

This discussion is to help provide an educational backdrop on what truly comprises end-to-end system latency.

This engineering interview & discussion follows-up our previous industry interviews with people like Tom Petersen of Intel and Bill Alverson of AMD, but now featuring Guillermo Siman of NVIDIA, who joins to discuss the engineering aspects of latency. We’ve recently been talking about an increase in our focus in testing on latency, adding to our existing FPS benchmarks, and this discussion is to help provide an educational backdrop on what truly comprises end-to-end system latency. We’ll focus on the latency pipeline, educational topics such as how the game engine handles and processes input, peripheral latency, and system-level latency (such as from the render queue). These topics also cover various types of bottlenecks.