Curiosity: A small Benchmark

Mephistopheles

Distinguished
Feb 10, 2003
2,444
0
19,780
Just out of curiosity...

...I wrote this small program in Free Pascal (as in <A HREF="http://www.freepascal.org" target="_new">http://www.freepascal.org</A>) that draws a 3D projection of a 2-dimensional wave with shading... It's primitive, but it's just a curiosity. And here's the thing: the drawing process takes a full 3.5 seconds in my computer, and the small program I wrote <i>counts time</i>, so it acts as a small benchmark. Here's the <A HREF="http://www.ifi.unicamp.br/~daviddc/bench.zip" target="_new">link to the program!</A>

So my time in this benchmark is around 3.48 - 3.52 seconds with a P3 933Mhz, 256MB PC-133 SDRAM, and a Maxtor 7200rpm ATA-100 40GB (heck, I forgot the model designation for this thing... whatever)... Can anyone give me any additional times? Just curious...

(too sad I can't get times below 1 second and make fully-fledged animations... It's all in software, anyway... And at 131K triangles, it would be even wiser to try implementing a "scanline rendering" algorithm.... It's all really primitive, and doesn't make use of any advanced graphics tech because... I'm too busy to learn to deal with them right now. Maybe when I've got more time and mess around with the code...)

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i><P ID="edit"><FONT SIZE=-1><EM>Edited by Mephistopheles on 05/24/04 03:35 PM.</EM></FONT></P>
 

Piccoro

Distinguished
Feb 4, 2004
70
0
18,630
copy/paste link in address bar

1.4 seconds
AXP @ 2.2GHz 200(400)FSB / 1GB (2x512MB) PC2700 2-2-2-5 / WD Raptor
This isn't completely accurate as there may have been apps running in the background. Should give you an idea though. I could run it from a fresh restart though (no apps running).<P ID="edit"><FONT SIZE=-1><EM>Edited by piccoro on 05/23/04 11:49 PM.</EM></FONT></P>
 

Mephistopheles

Distinguished
Feb 10, 2003
2,444
0
19,780
Ow damn, Yahoo doesn't support EXE files... And, while that was a zip file, Yahoo knew there was an exe inside. Very, very mean yahoo services... :frown:

So I'll have to wait until tomorrow to get on a nonrestricted server and put that file on the net. Sorry...

It'll be online soon...

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>
 

ChipDeath

Splendid
May 16, 2002
4,307
0
22,790
Just rename the .zip extension to something else, then Yahoo won't know it's a zip file. Although it does mean that we'd have to rename it after download, but that's not exactly a great hardship. :smile:

---
Epox 8RDA+ rev1.1 w/ Custom NB HS
XP1700+ @205x11 (~2.26Ghz), 1.575Vcore
2x256Mb Corsair PC3200LL 2-2-2-4
Sapphire 9800Pro 420/744
 

Mephistopheles

Distinguished
Feb 10, 2003
2,444
0
19,780
Well, I'm hosting the file at a University server at this<A HREF="http://www.ifi.unicamp.br/~daviddc/bench.zip" target="_new">adress</A>... It should work fine now!

It has been zipped, so you'll still need to unzip it... no renaming necessary!

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>
 

OrenSam

Distinguished
Feb 11, 2004
7
0
18,510
1.5-1.6 secs, specs in sig.

_____________________________________________
AXP 2500+, NF7-SL, 512MB DDR400, 9600XT 128MB
 

phial

Splendid
Oct 29, 2002
6,757
0
25,780
1.4 seconds @ AXP2.2ghz

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
 

P4Man

Distinguished
Feb 6, 2004
2,305
0
19,780
Euh, I just noticed disabling twinview (I use 2 CTRTs)increases the performance from 2.2 to 1.3 or so.. and changing my resolution from 1600x1200 to 800x600 makes an even bigger impact, so its not what you'd call a reliable benchmark :)

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 

jim552

Distinguished
May 1, 2003
171
0
18,680
On the following base system:
Asus SK8N Motherboard
1024MB RAM
ATI Rage 128 Pro with 32mb RAM

Opteron 242 processor 1.81-1.85
Opteron 246 processor 1.48-1.54

I may try to bring in a better video card and see if that makes a difference.

Already I know that any AMD processor over 1.4ghz won't score on performance tests any better on this video card than any other.

So I already know the video card can be a bottle neck on graphic opterations.

It's on a server, and My workstation so it's not too important.

Oh, and just for curiosity the Opteron 246 is abou 6.1 seconds if I run it across the terminal server!

Granted, NOT important, but I tried it.
 

jim552

Distinguished
May 1, 2003
171
0
18,680
I wouldn't contend that this would mean the reliability of the program is at fault?

It'a mainly for fun, but ANY change on a system can alter the results on ANY performance test.

In MOST current video cards lowering the screen colors to 16-bit will usually have an ADVSERSE affect as most video cards now-a-days are greated for 32-bit operation.

I wouldn't think this should be a surprise?

Changing the parameters under which the program runs, is just that. It has little to do with REALIABILITY.

In my case, just moving the mouse back and forth over the image being drawn adds about 20% to the time as well.
 

P4Man

Distinguished
Feb 6, 2004
2,305
0
19,780
Well, "reliable" may not have been the correct word, but clearly the results are meaningless without the proper context. I was under the impression this was a cpu benchmark, not a videocard benchmark, so if just changing the resolution or enabling a second monitor results in >100% changes, those numbers don't tell you a thing, especially not if people only report what CPU they used.

If I wrote a program that got the time from the RTC, displayed the last number of it and called it a benchmark, it would be perfectly reliable as well, just not very usefull.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
 

Mephistopheles

Distinguished
Feb 10, 2003
2,444
0
19,780
Nooo! It's not a videocard benchmark at all, it doesn't make use of any video card techniques.

However, P4Man, if you increase your screen resolution from 640x480 (307K pixels) to 1024x768 (786K pixels), you'll have to do more than twice the math......

So sorry, I forgot to mention that this is a highly <b>resolution-sensitive</b> benchmark. For it to be a CPU benchmark, you'd have to use the same resolution.... Forgot that....

I could write a program that uses only one specific resolution-output.... I really forgot 'bout that, sorry... I assume most people use the same resolution as I do - 1024x768.... Oooops....

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>
 

Mephistopheles

Distinguished
Feb 10, 2003
2,444
0
19,780
1600x1200 to 800x600 makes an even bigger impact,
Yes, that's four times less pixels (1/2 squared!)... As for 2 CTRTs, I suppose the program understands it has to render things twice, I don't really know about that though....

All projection and shading calculations are still CPU-based.... And if there was no output to screen, it would be a completely pure CPU and memory subsystem benchmark!

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>
 

gothitbycar

Distinguished
Dec 16, 2002
513
0
18,980
1.03 seconds, damn so close to under one. Going to retry with a small overclock.

A64 3200 2.0ghz
1 gb ddr pc 3200
2x36 gb 10000 rpm raid drives
Radeon 7000, 9800 pro is out right now

It's obvious the video card isn't a factor in this bench. Pretty neat little program you have.


-----------------------
[mind went blank]
 

Mephistopheles

Distinguished
Feb 10, 2003
2,444
0
19,780
That must be 9.85E-1, which is also known as 0.985...

Great score... What resolution are you using? That's an important factor (the only important video-card related factor! The rest is just CPU.)

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>
 

Ncogneto

Distinguished
Dec 31, 2007
2,355
53
19,870
There definatly are some inconsitancies here. Running a measly axp 2000+ I am get better times than some with 2500+'s and above.

2000+ @ 1024 x 768 = 1.19 sec
I do have 1 gig of Ram though

It's not what they tell you, its what they don't tell you!<P ID="edit"><FONT SIZE=-1><EM>Edited by ncogneto on 05/24/04 11:02 PM.</EM></FONT></P>
 

Mephistopheles

Distinguished
Feb 10, 2003
2,444
0
19,780
Then there has to be some difference. It's probably the only thing that changes, which is resolution! You're using 1024x768, which is probably less than what the others used.

So the "correct" way to indicate your score would be like you did it: AXP 2000+ @ 1024 x 768. This, for instance, compares directly to my P3 933Mhz @ 1024 x 768......

Come to think of it...

Now wait a minute... This isn't scanline rendering... then how come resolution seems to play such an important role? This is triangle rendering! It would seem that that is an intrinsic property of the triangle drawing procedures I used - which are part of Free Pascal. If this had been scanline rendering, then maybe 1600x1200 would be 4x slower than 800x600, but it's not!

Oh, and as for you having 1GB RAM: This plays a role alright, as well as the memory subsystem's speed. The 131K triangles used have, of course, 393K vertices with 3 coordinates each, so it's 1179K real numbers just for triangle position. Each triangle has its own normal vector, which is another 393K real numbers (3 coordinates for each vector, 131K vectors). Then there are, of course, the incident and reflected light vectors (another full 786K real numbers, all of which are temporary and can be wiped from memory quickly and replaced by the 131K values for shades... normal and reflected light vectors need to coexist in memory at some point, though). These all add up to about 1 to, at most, 2 million real numbers, which are flushed from memory at appropriate times, but I'm not sure I did it in the most intelligent and straightforward way for the poor memory controller.

Actually, the smartest way to do that would save up memory.... But bear in mind that I developed the algorithm itself in 2001, some 3 years ago... So give me a break.... :smile:

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>
 

Verikon

Distinguished
May 21, 2004
13
0
18,510
2.56 @ 1024x768

--------------------
P3 Coppermine 975mhz (150 * 6.5)
384 PC133
Geforce2 MX-400
Enermax 250 watt
Spray painted/modded P2 Dell case
<b>Cost: $0</b>
 

TRENDING THREADS