wgpu-native
wgpu-native copied to clipboard
Best way to do benchmarks
People like benchmarks, and often use FPS for this. This is not a great measure for several reasons. To name a few: it's nonlinear, at high FPS you're mostly measuring "overhead", and vsync and other rate limiting measures get in the way.
Are there ways and/or best practices to do benchmarks in a reasonably good way with wgpu-native? I suppose https://github.com/gfx-rs/wgpu/issues/1395 is relevant too.