miniB3D entity and/or sprite limit?
BlitzMax Forums/MiniB3D Module/miniB3D entity and/or sprite limit?
| ||
| what are estimated reasonable limits on the number of entitys and sprites before you would expect to see slowdown given the following assumptions: "average" hardware (i.e. not top of the line, and not 6+ years old. and not mac or windows specific) enough memory to handle everything without the system starting to page etc. relatively simple models (basic sphere or cube type model, not like a tree with tons of branches) how would these changes affect it. (a lot, a bit, a little, not at all) just using sprites, or just using the meshes with and without Collision detection (using updateworld) with and without 1 camera pick per frame I'm just curious if anyone has any opinions or experiences. I've managed to bog my system down with what I thought wasn't too many objects, but it usually turns out to be something exponential in my code that gets lose and not the engine at all. It also seems to me that my mesh size has much less to do with overloading things than the number of entitys I have. Yet the particle system examples I've seen all seem to use a ton of sprites, each one is an entity, though I thing I read somewhere there's a batching system for sprites so that's probably taking care of the entity count issue... Anywho, just curious about other's experiences, I'll do some tests and see what I find. |
| ||
| I did some experimenting and this is what I got in a general sense vertex count is the primary cause of slow down, not entity count creating 2000 cubes causes much less slow down than even 100 spheres with 64 segments entity count does cause some slow down though not much 2000 individual cubes runs slower than 2000 cubes all added to 1 mesh, but not drasticly textures cause exponentially more slow down than plain objects. this affects sprites too. sprites are slower than blank cubes entity alpha has no noticable effect on speed collision detection had no noticable effect on speed however no objects were moving let alone colliding I did this on a 3.2ghz 64x2 core windows box running vista x64 and an nvidia 8800gt, so it's hard to get accurate numbers due to os/core count/background load etc. lots of variables causing the fps to fluctuate wildly but I the general observations should hold true across other environments. |
   |