League Of Gamers FPU Benchmark program

[craigwhiteside]

ey guys here is a program i managed to make in my spare time, it is a Floating Point Calculation benchmark, i need people ready to test it to find out if it gives comparable results.

it operates on one core as default and you have to enable the other cores in task manager with set affinity.

im still working on the general layout and optimizing it, and will put dual core and quad core cpu selection for the benchmark.

Heres the forum link which leads to the benchmark :) it explains it a bit more

if your wondering its programmed in visualbasic.net using express 2008,

if you want to help be a developer for the program just pm me or go on our forums

http://forum.leagueofgamers.info/showthread.php?p=616#post616


Download Here

http://leagueofgamers.info/index.php?option=com_docman&task=doc_download&gid=4&Itemid=37

[craigwhiteside]

oh and try 1000000 instead of 1000 :)

[craigwhiteside]

anyone wanna test it :p

[JrRacinFan]

I am working with it right now. Doesn't seem too bad.

EDIT: Any possible way to do a validate or a C & P to do a comparison? Are you adding a database to your forums with results?

[craigwhiteside]

what score u get i get around 5 seconds

[JrRacinFan]

Around 9.2 but thats pretty obvious. You have twice a good proc as I do.

[pt]

http://img.techpowerup.org/071211/fpu.jpg :)

only using one core

[craigwhiteside]

yay it works for people :D

[technicks]

I got about 3.9.

http://img.techpowerup.org/071211/Capture007.jpg

[craigwhiteside]

nice score techniks, guess it does show which cpu is better :)

[craigwhiteside]

o.o cant see nothing o.O

[technicks]

Where? lol

[SpookyWillow]

heres mine on 1 core, x2 3800 @ 2.7

edit, couldn't get the image to work via tpu upload so had to host it myself :/

http://www.spookywillow.pwp.blueyonder.co.uk/bench.jpg

[JrRacinFan]

Dang that image took a long time to load. Try using imageshack or photobucket.... :p

[SpookyWillow]

instant for me :/

[pt]

instant for me aswell
ps: w1z is peeking :p

[craigwhiteside]

nice score spooky :D

[Jizzler]

Work box, P4D 3.4Ghz, 6.188 seconds.

[JrRacinFan]

instant for me aswell
ps: w1z is peeking :p

I figured out what was going on... I upped my DSL plan and they were working on fulfilling my order at the same time.

I do apologize Spooky, good time! Very quick.

[SpookyWillow]

thanks and no problem :P :D

[craigwhiteside]

ok i optimized the code a little bit, also added a bit of flavour to it :toast:

[craigwhiteside]

so guys what you think of this new one, tell me what scores you getting and see if its any improvement :)

[pt]

better score on both runs
http://img.techpowerup.org/071211/fpu532.jpg

:)
you need to do a database for it, and some sort of validation

[craigwhiteside]

ya, ill sort a database out, and put on our site :)

[craigwhiteside]

ok guys you can use this as a stability test too :), either set the iterations for 1,000,000 for about 5 seconds, 10,000,000 for 50 seconds, 100,000,000 for 500 seconds, etc
if you have multiple cores, just run the programs twice for dual cores and 4 for quads, and set affinities for all of them then run the test :)
once you done this all of the core's should be up to 100% load :)

[SpookyWillow]

ran the newer version and got this

4.250 seconds
time per 1000 iterations = .0043 secs

quite a bit of an improvement, a clean reboot was also involved too :D

[craigwhiteside]

hehe :)

[craigwhiteside]

updated your score spooky, the benchmark page is up :)
http://forums.techpowerup.com/showthread.php?t=46729

[DanTheBanjoman]

Like I said in your scorethread, I vote for making it multithreaded. Otherwise it's just another SuperPI or whatever single threaded bench. And since those are rather pointless nowadays it would be way more futureproof if it made use of the available processing power.

[DanTheBanjoman]

Let me try to be useful.

google found me http://www.devx.com/DevX/10MinuteSolution/20365/0/page/1

I believe your program does the same calculation over and over right? I'd say you can make an input field selecting the amount of threads. And jsut make that many threads, divide the iterations by this same number and have each thread do as many iterations as iterations/threads. At the end of each thread it could add +1 to some variable and when the variable is equal to the threads it's done.
Doesn't sound too hard in vb6, I doubt it's much different in .net.

Also I would increase the default amount of iterations or differences in time are gonna be really small.
I'm in no way a programmer or anything but I guess a basic example should be relatively easy to make. And then sample code can be replaced by your code.

Perhaps I'm a bit oversimplifying though when I have some time on my hands I'm willing to figure this out, does sound interesting.

[craigwhiteside]

ya ill try enable the multi threading, will decrease the times from around 50-75%, depending on whether you have a quad core or dual :)

[DanTheBanjoman]

ya ill try enable the multi threading, will decrease the times from around 50-75%, depending on whether you have a quad core or dual :)

Please don't limit to quad. Allow endless or at least 16 or so. Or you'll end up recoding within 2 years.

[tigger]

I got 2.9 seconds :)

http://img.techpowerup.org/071211/Untitled798.jpg

[craigwhiteside]

nice score :D
(adds score)

[craigwhiteside]

so guys what you think of it in general, like what would you rate it as, against similar fpu benchmarks, and you think it shows how fast a computer is compared to others?

[DanTheBanjoman]

so guys what you think of it in general, like what would you rate it as, against similar fpu benchmarks, and you think it shows how fast a computer is compared to others?

As long as it is single threaded my thoughts are the same as with SuperPI and the likes, it doesn't utilize the available processing power hence doesn't give a useful result. Besides, there is too little data to see how useful it will be. No clue what exactly it calculates. However I'll be following it, see what happens.

[craigwhiteside]

It determines execution speed on an extremely floating point, intensive real-world
application, it just checks accuracy on an algorithm that is exquisitely sensitive to errors. :cool:

the performance of my program is far more sensitive to changes in the efficiency of the trigonometric library routines than the average floating point program. :)

the trigonometric functions (also called circular functions) are functions of an angle. They are important in the study of triangles and modeling periodic phenomena, among many other applications :D

hope that clears it up :)

Edit: I know it sounds a bit complicated but once you know the theory behind programming its pretty simple :)

[spud107]

might chuck in the 4000 san diego n try on that. . .

[marsey99]

i get an error code, yadayadayada (0xc0000135) ????

[craigwhiteside]

i guess you havnt installed the .net framework :)

http://www.microsoft.com/downloads/details.aspx?FamilyID=10CC340B-F857-4A14-83F5-25634C3BF043&displaylang=en

[marsey99]

crap, that might be it :D

man their servers are slow, im getting it @ 204kps woohoo, its only going to take 2 days for 64mb :P

thanks for letting me know whats wrong btw.

[marsey99]

sorry bout the dp.

errm i cant realy see the text tho.

3.266 from 3.2ghz sound about right?

[craigwhiteside]

ya, seems as though people with vista can see it properly with a transparent background :shadedshu

[pt]

ya, seems as though people with vista can see it properly with a transparent background :shadedshu

xp ftw :D

[WOutZoR]

My score:

http://img.techpowerup.org/071212/FPU_Bench_resize.png (http://img.techpowerup.org/071212/FPU_Bench.png)

Ran the test a couple of times but this is the fastest time I got ;)

[craigwhiteside]

nice score WOutZoR :D

[Fizban]

Intel Core Duo 2 Merom T7200 I seem to average right around the 5.45 second mark.

[Fizban]

Set Affinity to 1 and with a single core I get right under 7 seconds.

[craigwhiteside]

nice score for a laptop :cool:

[Fizban]

Best score so far on the laptop:

[craigwhiteside]

fiz, try to use thumbnails instead of inserting it directly into the forum please :)

But nice score anyway :D

[Fizban]

I changed it to a thumbnail immediately after I hit submit and saw how large it was heh.

[Fizban]

I'm thinking the # of iterations needs to be a long or double, not an int because it won't let you do more than 2,147,483,648 iterations at the moment.

[craigwhiteside]

i didnt both making it a long cos i didnt think people would go beyond the max number :laugh:

but ill add it nevertheless :)

Dim iteration1 As Long

:D

[Fizban]

To ensure I stay the top laptop in this test....overclocked to 2.4 GHz and scored 4.352 seconds

[craigwhiteside]

updated :D

[snuif09]

My score with sempron 3400+ AM2 at 2.4 GHZ