This question already has an answer here:

I'm developing a voxel game with OpenGL and want to optimize the number of rendered faces. How can I estimate how many faces at most a modern graphics card of a certain brand is able to compute, while still managing to render at 60fps?


marked as duplicate by Anko, Kromster says support Monica, MichaelHouse Dec 30 '14 at 19:02

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.


There's only one sure way - You need to test it yourself.

Estimation result largely depends on your actual code (how efficient is it) and shaders (how complicated are they).

Also, there's number of parameters you need to take into account, each can dramatically change the estimation:

  • Antialiasing (ranged from Off to 16x ?)
  • Display resolution (0.5k to 4k)
  • Anisotropic filtering (ranged from Off to 32x ?)
  • other GPU settings (performance vs. quality slider)
  • slow CPU can also be a bottleneck

EDIT: Generally you need a bunch of testers, who will play your Alpha/Beta and from whom you collect performance data and decide if you need to reduce polycount or not. Also it is generally a good idea to add Options slider to allow players to tradeoff picture quality for speed. Some prefer higher FPS, some prefer picture quality.

  • \$\begingroup\$ Okay, but I can't really test my game on every single graphics hardware out there. But still I think you are basically right. So I will test it on a generation of GPU (say e.g. AMD Tahiti) and then extrapolate according to the different clock rates. \$\endgroup\$ – Entrusc Dec 28 '14 at 14:36
  • \$\begingroup\$ Added to answer \$\endgroup\$ – Kromster says support Monica Dec 28 '14 at 21:10

Not the answer you're looking for? Browse other questions tagged or ask your own question.