Parhelia Reef Demo & FRAPS

[Guru]

Originally posted by az
Thop: In the case of the fr-08 and fr-19 demos, one of the reasons surely is that space was one of the main concrens, and simple techno "music" can easily be done by a midi-like system, which is then sampled in real-time. Another reason would be that it kinda fits the trippy looks of those demos. Oh, and computer geeks have no taste

AZ

fr-08 used 800MB of audio data if I rememberd the end text correctly!

[Maggi]

Originally posted by Guru


fr-08 used 800MB of audio data if I rememberd the end text correctly!

I think the keyword would rather be, they generated 800MB of raw audio data, then add the visuals to that and be stunned how they packed it all into 64KB !!!

[GNEP]

Basically, they are just astounding.

This reminds me of Braben and Bell's original Elite on the BBC - very tight coding, which (iirc) generated galaxies using some sort of algorithm and a non-random seed. V. impressive.

I'd guess the textures here are being generated form fractals somehow?

gnep

[CHHAS]

Uhmmm.... Elite

[GNEP]

Have a look at www.newkind.co.uk for a bit of retro action... this brought back loads of memories!

gnep

[CaineTanathos]

ontopic

my reefdemo +fraps results :
Minimum FPS : 25
Maximum FPS : 51
Average FPS : 34.5
STdev:5.7


Specs :
P4 1.8@2.4
1x 512 MB PC2700 @ 187 mhz
Parhelia 128R 1.0.4.231

[CaineTanathos]

ok last results where only one loop

these are the results running it 30 mins

Minimum FPS : 24
Maximum FPS : 49
Average FPS : 32.96
STdev:5.17

[GNEP]

Hmmm... those are a bit lower than my t'bird. Interesting. But seems that the CPU is not a huge factor here.

I have (since the original numbers) managed to push my t-bird a bit higher. The multiplier is still locked (at the wrong setting it would seem), but with my crappy mobo I tracked down a BIOS that allows me to change the FSB. So now running at 1.265 GHz (FSB@110), up 10% on before. Looping the demo several times (10 or so), and separating the (remarkably consistent) loops out, in order to only average over complete loops, I get the following numbers:

Min: 25 (the very first one was 23, but I discounted it as it was right at the beginning of the first loop)
Max: 54
Avg: 37.4
St Dev: 6.2

The increase in clock appears to have raised the peaks more than the troughs

gnep

[TdB]

I just tried to run the reefdemo with that 3d-analyse utility, apart from a fps-counter it also has polygon-per frame-counter and an polygon per seconds-counter.
I must say the poly-count in the reefdemo is HUUUGE
it is varying between 70.000 and 180.000 polies, and it renders between 2M and 4M polies per second.

that might not sound like alot, compared to the highpoly test in 3dmark2001, but renember the these polygons are multitextured and pixelshaded/bumpmapped, and it still has the power to execute some pretty advanced vertexshader programs AND keep the average framerate above 30fps.

Im impressed.

[GNEP]

Where can we get this "3d Analyse" program?

Thanks

gnep

[TdB]

http://www.tommti-systems.de/start.html

in the "files" section.

[GNEP]

Cool. Cheers TDB.

[CaineTanathos]

Originally posted by GNEP
Hmmm... those are a bit lower than my t'bird. Interesting. But seems that the CPU is not a huge factor here.

I have (since the original numbers) managed to push my t-bird a bit higher. The multiplier is still locked (at the wrong setting it would seem), but with my crappy mobo I tracked down a BIOS that allows me to change the FSB. So now running at 1.265 GHz (FSB@110), up 10% on before. Looping the demo several times (10 or so), and separating the (remarkably consistent) loops out, in order to only average over complete loops, I get the following numbers:

Min: 25 (the very first one was 23, but I discounted it as it was right at the beginning of the first loop)
Max: 54
Avg: 37.4
St Dev: 6.2

The increase in clock appears to have raised the peaks more than the troughs

gnep

I was thinking (and yes it hurts like hell ), maybe the drivers aren't optimized yet for the different cpus. And since the AMD has a better floating point unit than the standard of the P4 its faster.
But I am not sure if thats the reason. Maybe its normal

Which drivers did you use?

I am gonna increase the agp busspeed, see what happens

[GNEP]

I use the public 231 drivers. I also get lower 3dMarks than you (about 6150). When I am back with my machine towards the end of the month, we should do some comparisons with real games!