Posts: 693
Threads: 186
Joined: Jan 2018
Reputation:
1
Its so much more important than people realise. I cover it a little in my book, but here's a really great video to see it in action, this shows how you can make a small amount of gpu go a long way
Brian Beuken
Lecturer in Game Programming at Breda University of Applied Sciences.
Author of The Fundamentals of C/C++ Game Programming: Using Target-based Development on SBC's
Posts: 693
Threads: 186
Joined: Jan 2018
Reputation:
1
aww the video is gone, I'll try to find it
Brian Beuken
Lecturer in Game Programming at Breda University of Applied Sciences.
Author of The Fundamentals of C/C++ Game Programming: Using Target-based Development on SBC's
Posts: 693
Threads: 186
Joined: Jan 2018
Reputation:
1
12-09-2019, 11:13 AM
(This post was last modified: 12-09-2019, 11:35 AM by Brian Beuken.)
not the same as the original video, which was awesome, but here's the principle again in a shorter video
https://www.youtube.com/watch?v=BgQZo-_282g
There are several methods for culling things, frustrum and occlusion being the most widely used. The basic principle is that while the GPU is much much faster at drawing things than the CPU, so you can often let it waste time drawing things not seen, a little time spent by the CPU or GPU to determine if an object should be drawn while not visible, can spare a massive number of GPU cycles. Doing a thousand (even hundreds of thousands) tests before attempting to draw a model or terrain section, can make a massive difference to performance and have no effect at all on the visible result. This is especially true in terrain or environment based games where you are in a 360x360x360 world populated by much more than you can see at any given time.
Brian Beuken
Lecturer in Game Programming at Breda University of Applied Sciences.
Author of The Fundamentals of C/C++ Game Programming: Using Target-based Development on SBC's