Warzone 2100 Forums

Tested on rpi 3b+
Could be any compilation flag or configuration improve performance on arm boards??
The pi its running at 1200 mhz because of heat. For sure it can run beter with overclock and active cooling. The bottleneck its clearly on cpu because any resolution changes the performance, not even antialias.
Anyway... Great game guys.



https://youtu.be/0owiI0d28Mc

I'm actually amazed it's running that well on an rpi3 b+. (What OS are you running?)

Make sure you are compiling it in Release mode with all optimizations enabled? (If you aren't, you'll definitely see much higher CPU usage.)

Not sure if this may be helpful: https://www.raspberrypi.org/forums/view ... p?t=223592
(Warzone currently supports only desktop OpenGL, not OpenGL ES.)

Mesa provides full opengl api...sorta. Thats why the pi could handle the game. As I said. Gpu power its not the problem here i need to check the cpu side here. Thanks... It was really easy to compile.

It is raspbian. But on my channel i test different mesa drivers. Currently on mesa 19.has some pros and cons. But as you see. Having a full opengl api open all the doors.

Mmmm... It only uses one core. Thats the problem. And you cant fix it without rewrite the all thing. A shame. But good performance to only use one core.

Maybe it you can provide some dummy multi thread support with something like these (founded on this forum)

CORE 0 - Main/Critical code, Player
CORE 1 - path-finding calculations
CORE 2 - AI Players, scripts, events, triggers
CORE 3 - something else (mb net code)

I belive the pi3 its power enought to blow your expectations running on multicore.

It already runs path-finding on a separate thread, which your system should migrate to a different core as needed. It is probably not ideal, in that we should rather send paths over the network rather each host calculating every path, as we do now - but there are advantages to both methods. We could somewhat easily run multiple AIs/scripts in parallel as well, but we would have to wait on the slowest one, and batch up actions to run after, so it would not always be faster (probably make triggers that call into only 1 script significantly slower). Most other stuff would be hard to make parallel since the synchronization overhead would probably kill any performance advantage that could be gained from it.

Ok. Too much for my poorly pi hahahwha. She does what she can!! Anyway... The usage of cpu its very low. 20%. Thanos for the support. Ive never played the game, I am just a tester but some people ask mem to try to improve it a bit. I checked the release mode and add some rpi flags but the thing doesnt change too much.

Good Information..very useful for my studies