You'll need to make the logic framerate independent from the rendering framerate. This will make it irrelevent how many rendering FPS you are getting.

I usually clamp my logic FPS to 30, and let the rendering go wild.

This topic has been run over many times on the forums. Try searching the forums for "rendering logic independent".

The number of times the scene is rendered/displayed to the user should NOT affect how the games is played assuming you can render the scene >= 30 times a second.

Just use your timer system to make sure your game logic doens't get updated more than once every 33 milliseconds.

Quote:Original post by ma_hty

Quote:Original post by dgreen02
...
I usually clamp my logic FPS to 30, and let the rendering go wild.
...

Rendering more than 60 fps is not very useful as the refresh rate of the monitor is usually 60 fps. It just do nothing besides keeping the CPU busy.

However, the input device have a far far higher refresh rate, 30 Hz for game logic is too slow a respond.

I think... the converse should be better, i.e. clamp rendering FPS to 30, and let the logic go wild. Just make sure you are using the system timer to tick the progress, it should be fine.

Either way you're going to chew up the CPU, why leave it idle if your game is running fullscreen? Of course there's no point in rendering more than 60 times a second, nobody will notice it, though it's a nice way to gauge things. Every engine does it, unless you turn vsync on to get rid of the tearing.

I say to clamp the logic @ 30 times a second because I'm used to complex scenes and logic updates where I'm lucky if I can get 30FPS on a high end machine. I'm not sure what the OP has in mind for the future, but in a complex physics scene xxx,xxx collision triangles and 1000s of actors you'll be heavily CPU/logic limited.

I suppose for a 2D game or lite 3D game or something where the controls might matter to that degree, yea you should put the focus on the logic...but I can't tell a difference personally.


Anyways the solution to his issues lie in making the logic/rendering independent of eachother using a timer system. Once that's in there he can clamp either at whatever rate works best for his setup.

Here are two articles on the topic right here at gamedev if the OP hasn't already found them:

Achieving Frame Rate Independent Game Movement

Frame Rate Independent Movement

Quote:Original post by ma_hty

Quote:Original post by dgreen02
I say to clamp the logic @ 30 times a second because I'm used to complex scenes and logic updates where I'm lucky if I can get 30FPS on a high end machine. I'm not sure what the OP has in mind for the future, but in a complex physics scene xxx,xxx collision triangles and 1000s of actors you'll be heavily CPU/logic limited.

I'm sorry. I don't quite follow your logic.

Yes, although the game logic is usually less demanding than the graphics routines, it can happen to be more demanding. However, if it does happen this way, why would it help to force the game logic to perform at a lower rate explicitly?

For me I don't notice any difference + 30 frames per second in terms of the game "logic" updates, and clamp it at 30 so I can make sure things run the same on every computer...which infact would solve the user's issue above. When it comes to stability of physics engines and the 1000s of variables that can be tweaked it's nice to limit the range of possible updates-per-second, for instance. Also when doing multiplayer stuff, it's nice to limit the # of updates for certain things.

Also you have to understand that "setup" ie: building vertex buffers, potential visiblity, etc. all fall under the catagory of "logic" and not "graphics" and in what I was describing...in very complex scenes the physics and this other stuff will take up more time than rendering, it all depends on what you're doing.

So you update your logic 30 times a second, render 60+ times a second...lerping between previous frames to provide smoother visuals, and using less CPU time.

If you are limited by the logic and let them both run un-checked you'll end up somewhere in-between at 45 FPS with mixed results on all types of hardware [ cpu/gpu combinations].

I find that computers these days [ say an Intel 6600 + Geforce 8800GTS ] are going to be CPU limited in complex games/applications.