Noizze

2nd February 2002 06:54 UTC

Benchmark
Everything computers has a benchmark. I have not seen one done for AVS though. So I am proposing that we decide on a set of presets and size of the AVS window and other various settings. I am not sure if AVS uses integer, floating point or both. But these kind of things are what should be concidered when deciding on a preset.
So is anyone interested? If this has already been done please tell me so I can shut up, and go back to my corner.

UnConeD

2nd February 2002 16:51 UTC

AVS is influenced by many things... you'd need to settle on a standard MP3 too, as CPU usage can vary if the currently playing MP3 is a complex or simple one. Not to mention the fact that AVS's FPS read-out is all you could benchmark, which is not very exact.

Noizze

2nd February 2002 19:42 UTC

Understood, but a audio source is not required for avs to work, just not as exciting. The FPS may not be exact but, appears to be stable in its readings. For example:

Preset= UnConed - Energy Box
Audio= None
Size= 320 X 240 Windowed
Windowed performance= Max
FPS= 22.2 Overlay= Off
FPS= 19.8 Overlay= On Set desktop color= off
FPS= 19.4 Overlay= On Set desktop color= on
All other check boxes= Off

Size= 640 X 480 Windowed
Windowed performance= Max
FPS= 10.5 Overlay= Off
FPS= 8.8 Overlay= On Set desktop color= off
FPS= 8.1 Overlay= On Set desktop color= on
All other check boxes= Off


OS= Win98 with 98lite
CPU= Celeron1 366@550 5.5X100
Chipset= BX440
VideoCard= Matrox G450 32MBDDR Dual head, Overlay active

This is an example, the numbers are correct, but just a sample.:D

It is interesting to note that while to # of pixels quadrupled the FPS only decreased by around 45%, not to >25% as one might think.
BTW UnConeD I love your work!

Montana

3rd February 2002 11:50 UTC

to noizze:
some avs doesn't do anything without music
so don't say that avs is independent

UnConeD

3rd February 2002 22:08 UTC

Indeed... I believe some AVS effects are optimzed so that they won't do anything useless. So blurring a completely black image (or partially) will be faster than blurring a regular one. You could always use the tone generator or something :)

solamon

4th February 2002 20:33 UTC

I agree with you noizze. But wouldn't it be easier just to benchmark the CPU of your computer seeing that AVS only uses the CPU?