What Most Effects Game Engine Performance

That’s my question. I typed this once but the forum assed out on me so plz forgive for bluntness now:)

I know that physics cost you a lot of performance. And I have stress tested my PC to run 22,000 polygons @ 60fps with materials and UV maps.

However I don’t know how much of an effect armatures and animation, and logic and scripting will have on it - seeing as I have not enough of these things to really test it out. There are other processes too, like referencing disperate scenes and resources that aren’t easy for me to test out right now. I’ve searched online and haven’t foudn any stats either.

In other words - having stress tested 22,000 polys I need to know whether or not animation could have a great affect on my overall poly count. I would hate to think I have 22,000 polys to work with when I only have 10,000.

If you can help shed some light on things, thank you.

Good question there mate :wink: Hmmm… Well I’m a newbie at the area, so I cant really help you out, but if I figure anything out I’ll let you know :slight_smile:

The polycount and texture size are the main things you need to keep an eye on IMO.

The logic and other processes are usually not the source of the “framerate funnel”. It’s almost always renderer oriented when talking about performance issues.

As for the “official” stats, I don’t think there is a bar set anywhere, since the BGE’s performace pretty much depends on your hardware, so it’s different for everybody.

I’m trying to use as few textures as possible as a matter of fact thanks

I think it would be interesting to take 30 objects and apply the same sensor to each of the 30 objects. Bench it and then try a new sensor and repeat then bump up the number of objects. Or similiary apply a (single armature to an object )X30 and then sensors and bench the two

So we could see when armatures are worse. Or how many armature break your system down. If we recieved similar break down points across varying computers it would show a problem in blender. Or from above rank the demand for each sensor relative to the other sensors. always sensors = .3(Near sensors)*(x^n/y^m)

I tried that already, but what I haven’t tried is actual armatures with real movements and recorded actions and stuff because I haven’t finished a whole lot of that kind of stuff - I duplicated an armatured 1000 poly mesh that was uv mapped and animated by an experienced person and I duplicated it about 20 times and set sensors on it so that it can spin and it that’s where I got my 60 fps at, if I added any more objects it began to slow down substantially.

I just needed to know if there was some other thing I didn’t know about that would slow it down more so I can take that into account for models