"Vision without action is a daydream;
action without vision is a nightmare."
- Japanese proverb
More pages: 1 ... 11 ... 21 ... 31 ... 41 ... 51 ... 61 ... 71 ... 81 ... 91 ... 101 ... 111 ... 121 ... 131 ... 141 ... 151 ... 161 ... 171 ... 181 ... 191 ... 201 ... 211 ... 221 ... 231 ... 241 ... 251 ... 261 ... 271 ... 281 ... 291 ... 301 ... 311 ... 321 ... 331 332 333 334 335 336 337 338 339 340 341 ... 351 ... 361 ... 371 ... 381 ... 391 ... 401 ... 411 ... 421 ... 431 ... 438
Query FailedWester
Sunday, August 22, 2004

Er I meant
Wester547
Sunday, August 22, 2004

Hey Humus,

Very nice work.

Running an ATi Radeon 9500 Pro here, DirectX 9.0c and Catalyst 4.8 drivers. When I tried the "each particle draw per call" option, my frame-rate slowed down to 10-30FPS. Instancing, 150-300FPS, Using the Vertex Shader Constant Instancing method, 100-150FPS. And the buffer method, 60-90FPS.

Impressive use of the Instancing / Vertex Shader 2.0 specification techniques and [volumetric?] particle effects, keep up the great work man! ;-)

Wester547
MesserFuerFrauSchmid
Saturday, August 21, 2004

Why don't you all shut the crap with instancing does not work blaaaah. I think people come to this conclusion since none of you have ever coded a real 3D application and not just a fancy looking effect or demo. One of the most often found bottlenecks is the batching of draw calls (of course they are CPU dependent since they go through system->API->driver->card). Instancing can exactly help to reduce this batching crap.
Advman
Saturday, August 21, 2004

Humus, your conclusions seem to be true. I've a 6800GT, default clocks, and I get completely different numbers than everybody else with metho 2 yielding highest fps, thus we are limited by CPU, memory, whatever, which directly leads to the conclusion, that instancing can't matter too much in the whole picture. Anyway, love your work!
Humus
Saturday, August 21, 2004

BetrayerX,
it doesn't fluctuate much for me. It's within +/- 5-10 fps or so, which isn't much when we're talking about 400fps. Good enough to get a sense of it. The problem though is that the first three methods are all quite fast, so they all are very CPU dependent. I get only a little better performance at home (X800pro) than at work (Mobility 9700), so it's not very useful anyway to compare between cards.

Jecht,
yes, it should work.
Utrion
Friday, August 20, 2004

a really nice upgrade for all of your demos would be a benchmarking utilitycause all of the demos are really good looking and have nice technics
Jecht
Friday, August 20, 2004

This demo works with Cat 4.9b, right?
BetrayerX
Friday, August 20, 2004

Nice work Humus!
For the people posting scores....don't bother. I have a 9700Pro and my numbers beat yours big time (429fps on mode 1 to set an example).... And we all know that the idea of a 9700 beating a 6800 is preposterus.

The numbers fluctuate so much that it is simply not accurate. @Humus...Plz if possible do a bench mode....dunno, camera fixed for 20 seconds in each mode would be nice:

Anyway, I am amazed that some new tricks can be squeezed on this old dog....NICE work HUMUS!
More pages: 1 ... 11 ... 21 ... 31 ... 41 ... 51 ... 61 ... 71 ... 81 ... 91 ... 101 ... 111 ... 121 ... 131 ... 141 ... 151 ... 161 ... 171 ... 181 ... 191 ... 201 ... 211 ... 221 ... 231 ... 241 ... 251 ... 261 ... 271 ... 281 ... 291 ... 301 ... 311 ... 321 ... 331 332 333 334 335 336 337 338 339 340 341 ... 351 ... 361 ... 371 ... 381 ... 391 ... 401 ... 411 ... 421 ... 431 ... 438