CPU Math Mark Scores

[Bhunta]

Im looking to see what other people get in high end AMD systems
You may have seen the program at tweakfiles lately, but I've been getting little or no high end amd score feedback. (I am one of the two authors of it)
you can download it here (http://www.tweakfiles.com/cgi-bin/dl-ftp.cgi?ftp.zdnet.com/threedee/tweakfiles/benchmark/CPUmathMark.exe)
my 112x8.5 (952MHz) Duron gets 57.3 seconds

[Hoot]

I'll start the ball rolling on this. My rig got 36.082 seconds.

If this has a loop mode, it would also be an excellent burn-in benchmark. Forget all that distirbuted computing philanthropy. It gets your CPU up there and fast!

Hoot

[Hoot]

Hoot (May 26, 2001 09:45 p.m.):
I'll start the ball rolling on this. My rig got 36.082 seconds.
Hoot

Whoops, that was at 10x150 (1500 Mhz). At 10.5x145 (1524 Mhz), it's 35.304 seconds. Wait a minute! Did I just confess to lowering my FSB to get a higher overall speed? Oh Dear, I feel like such a pariah. ( sic ) ;D

Hoot

[Bhunta]

That is the fastest I ever saw that run
do you think you could attach a screenshot so I can get your system info (so I can add you as a baseline).
I attached my whopping score...
a looping option is a cool idea.. I'll be sure to throw it in the next version :)

[wildone]

here is just as a reference but the system used for comparison must have had something dead cause I was just checking out the prog. and had all sorts of stuff running and I beat there score , can't wait till my new bios gets here and the old t-bird is back up and running.
Wildone

[illKainYurAss]

Hate to break it to ya Wildone.. but Higher is bad, lower is better.
My Sacky system:

AMD K6-2 500
Sack of !^&@#%!^# Motherboard ( DTK ) WTF is DTK?? <---- Source of crappiness
192 mb of PC100, 1-2-2
Asus Geforce 2 MX 32mb
Win 2k SP2, DX 8a

Visit my web page (http://technogarb.multimania.com)

[wildone]

Hate to break it to ya Wildone.. but Higher is bad, lower is better.
My Sacky system:
AMD K6-2 500
Sack of !^&@#%!^# Motherboard ( DTK ) WTF is DTK?? <---- Source of crappiness
192 mb of PC100, 1-2-2
Asus Geforce 2 MX 32mb
Win 2k SP2, DX 8a


I know thats what I am saying you beat the reference system too
see on my post where I had the k6-2 reference system,I beat it by some seconds
Wildone

[Hoot]

That's why they call this Overclockers.com ;D

Hoot

[Pitspawn]

Where can I find CPU Math?

[Bhunta]

Pitspawn:
download CPU Math Mark here (http://www.tweakfiles.com/cgi-bin/dl-ftp.cgi?ftp.zdnet.com/threedee/tweakfiles/benchmark/CPUmathMark.exe)
be sure to post your score :)

nice hoot :)
1525MHz
hehheh
I'll add you in as a baseline

[WinXP]

[img="http://albums.photopoint.com/j/View?u=1675561&a=12850577&p=49250609"]

Don't know how it rates but here it is.

[WinXP]

.

[Hoot]

Fresh boot up, no TSRs like MS Office Fastfind, MBM, CreateCD, etc running in the background. All TSRs steal your performance, litle by little. Turn off the modem. You should be cleaning my clock running 130 Mhz faster than me.

Hoot

[vipez888]

Hey on my rig, AMD 1.2ghz 200FSB AXIA@1.33ghz,

It managed to complete that test in 40.338s

Too bad I haven't tweaked my system enough yet, I've loved to see 3X seconds myself.

[Bhunta]

hey WinXP
Your score may be a little low do to a few things
It could be your motherboard.. some can transfer data at higher rates than others...
It may also be your bios settings. By disabling things in your bios, your score can be affected greatly. If you have any other programs running during the test, this can affect it too. Im not sure what else you could try.. Windows 2000 scores are anywhere from 2%-10% faster because they optimize your cpu better than 98.
There are a number of things your can try.. if you find out whats wrong, let me know :)

[WinXP]

Well I end task of evryting in background running except explorer an disable agp fastwrite, changed to 2x agp an changed to 2 way interleave. This is 11x155 cas2 setting.

[img="http://albums.photopoint.com/j/View?u=1675561&a=12850577&p=49261364"]

[Bhunta]

well regardless you beat my duron 700@955 :)

[oc jason]

my score

[castle lager fan]

Here's my score for your reference.

[Bhunta]

Likuid fusion.. is that a Tbird core?
nice score castle lager fan... the record is 35.4 seconds set by hoot :)
your within 2 seconds...

[Bhunta]

Check out the Hall of fame (http://www.31337.s5.com/halloffame.htm) I just set up

[WinXP]

Bhunta---- Thats a nice score you have there. What tweaks did you use to get that low ?

[Bhunta]

I have the duron 700@955MHz. I get 57.4.. you beat me
If you mean it's high for my clock speed, it was at 112x8.5 with completely optomized bios settings. (CAS 2, Turbo, Fast CPU command decode blah blah blah).
I don't really know what else to say. I don't have anything running when I get that score, and I have the latest updates/service packs for windows 2000 & via drivers. By the way what motherboard do you have? Some motherboards can transfer data etc. faster than others.. I have an ABIT KT7-RAID. Tell me if you find out anything :)

[WinXP]

I am using a IWill KK266R but not setup for raid. Cpu-T-bird 1.33 with everything at the aggressive setting. My lowest score was lowering some of the settings, which doesn't seem right. But I will play with it a little more an see if I can get lower.

[TT120]

Here's mine:

[Hoot]

The water is lapping at my feet! :-D
Way to go TT120.

[TT120]

Thanks Hoot.
Here's one with all the TSR's turned off. Beat ya by a smidge. :):):)

[WinXP]

Nice score there TT120, looks like I need to put my KT7A Raid back in an see how low I can get. I am missing some setting with this IWill board, can't get mine any lower..

[vipez888]

Hey here's my rig @ 1.4ghz from 1.2ghz, not stable but stable enough to pass this CPU Math mark, shaved off 2 secs from last try, hehe.

[TranceBear]

Well I had to tune down my system for the moment, but even back at 1.33, I think I get an impressive score.

[TranceBear]

Well this may be the new way to see who is the fastest, I twaeked a little and got it to 40.885. still running at a lowered speed until my new fan/heatsink comes in.

[Big Mike]

Here's mine, a few seconds off our leaderboard but I'm happy with it...

[Bhunta]

those are some nice scores :)

looks like there is a close battle between TT120 and Hoot :)

[Hoot]

So, where are all those 1.6, 1.7, 1.8G peltiered and water cooled rigs I see posts about from time to time. Seriously, I expected to hold first place for all of a day or so. C'mon "Heavy Hitters" put me to shame.

Hoot

[TT120]

You dont have first place anymore Hoot (grin) but second aint bad. :):):)

[Hoot]

TT120
I didn't see your 2nd shot. Congrats. Time to get out the "Freeze Spray" ;D

Hoot

[Tomas]

my result was 42.331 seconds. Is that ok speed for an 1.1@1323mhz?? (266)
altought i had an irc server and 2 mirc clients running in the backround but dont think they used much cpu power.

mobo=asus a7v133a 320mb ram

[Tomas]

41.64 now

[asmodean]

Here's my system. Check the sig for more specs.
The program looks REALLY cool, I like the titlebar. Maybe I'll do a Sawfish theme based on it, if it's okay with Bhunta? Also, does the program abort itself, if there's an error in any of the calculations? It should, otherwise people could run non-stable systems and get better results than with their normal config.

[Bhunta]

hey asmodean :)
I did the programming, the actual program GUI was done by vorte-x- (more info in readme)
Your welcome to do whatever you want with it for your own purposes (and your friends as long as you don't release it or anything).
I was thinking of adding a burn-in feature that just runs the program in a loop a specified number of times
your idea of a calculation error is cool.. but I don't know how much control I could have with it (lock-ups etc. are hard to stop)
anyway if you have an idea, drop me a line, thanks for the feedback!

[WinXP]

TT 120----- What bios are you running? Have you changed any of your settings with wpcrest? Just curious I reinstalled the KT7A Raid today an still can't get less than 41. What version of Windows are you running?

[TT120]

WinXP (May 28, 2001 04:15 p.m.):
TT 120----- What bios are you running? Have you changed any of your settings with wpcrest? Just curious I reinstalled the KT7A Raid today an still can't get less than 41. What version of Windows are you running?I'm running the ZT rev of the bios. I have never heard of 'wpcrest' so I guess the answer to that one is no :) I did, however, turn off all my TSR programs to get it that low. My first score with all the TSR's that I normally run all the time was in my first post, about 35. 6 or something like that. I'm running Windows 98se.

[oc jason]

Bhunta (May 27, 2001 06:08 p.m.):
Likuid fusion.. is that a Tbird core?
nice score castle lager fan... the record is 35.4 seconds set by hoot :)
your within 2 seconds...


Yes its a tbird core why do you ask-its a slot tbird though, maybe that explains what u wanna know-and the ram is pc100-system is not overclocked yet

[Bhunta]

thanks :)
don't know if I'll use it yet, but I might use a couple in this thread for the next release baselines

[WinXP]

TT120--- TSR ? Is this what is running in the backgound? I only have explorer running in the background when I run the program, everything else is closed. I am running the ZT 1.11+ bios with Win98se also.Can you get your mobo pass 147fsb at cas2? Mine die's there an will only goto 150 at cas3. This is with Corsair PC150 cas2 256meg stick.

[Hoot]

Just finished major mod to the Swiftech MC-462A. Shaved almost 5C off of the loaded temp!
"Here's Johnny!"

Hoot

[TT120]

WinXP (May 28, 2001 10:03 p.m.):
TT120--- TSR ? Is this what is running in the backgound? I only have explorer running in the background when I run the program, everything else is closed. I am running the ZT 1.11+ bios with Win98se also.Can you get your mobo pass 147fsb at cas2? Mine die's there an will only goto 150 at cas3. This is with Corsair PC150 cas2 256meg stick.Yeah TSR's is whats running in the background. If you press CNTL-ALT-DEL, you'll see what you have running as TSR's. You can run msconfig and shut all that stuff off. I run mine at 148 FSB cas 2 but thats as high as I can get it. It gets unstable after that. I can go higher but it dont run too good up there. 148 is solid as a rock.

[TT120]

Good job Hoot! Heh, guess I'm gonna havta bow down to the fastest. I cant get mine to run stable any higher than what I have it at right now.

[castle lager fan]

Good job Hoot.

Maybe you can give me cooling lessons on the MC462. I'm only on air and will stay that way.

I could have got a better score but need better mem, maybe the nes 150 MHz Samsung DDR with 2 CAS. Mine is only 133 DDR with 2.5 CAS, which I run with 2 CAS. I had to go back to 2.5 CAS for the score with 138 FSB x11.5. The CPU got too hot with 12 x 138 50C + at idle too quickly to my liking so I stopped and went back. Maybe another day I will break 1.6 GHz plus. Anyhow here she is!

[castle lager fan]

Sorry double post.

[Hoot]

Yo, Beer Man!

Hoot

[asmodean]

Erhm... could you people use PNGs instead of JPGs? They take less space when palettized and are lossless. Just a thought...

[TranceBear]

Well my new fan and heatsink came today. I am back up to 1.46. OH YEA! My new score is nice.

[Bhunta]

your #4 Luv2Rave
you better watch it though
my duron will eat you alive :)

[castle lager fan]

Hi Hoot,

How do you keep your CPU so cool? I keep going to 45-46 C on mine with the swiftech 462 and Delta fan.

Are you going to try for more?

By the way my motherboard is a FIC AD11.

Where's all the big guns with 1.6 GHz plus? I would like to see how fast those machines go, as I'm unable to break 1.6 GHz yet.

PS : Hoot I'm back at 1.466 GHz as my CPU get's too HOT!! (40C Idle 46C Load Prime 95). My case is also a hot house.

[FNG]

AXIA Y 1.33@1.33
Asus A7M266
256 MB of PC2100

[Danny]

24.88 sec. Complex

38.797 Over all

1.4GHz Tbird (140x10), DDR @280MHz CAS2, AMD760

[Hugo 59]

Here's another for you.

[asmodean]

This picture shows the power of AMD, when doing math... The box is from work.

[asmodean]

Here's another one, demostrating the difference between w98 & NT... It's intel, but it shows a nifty difference. Couldn't find the CPU Math Mark thread in the intel section, so I'll post it here.

[Bhunta]

I ran the celeron 500 test on these computers at school with the cheapest motherboards money can buy in them. I guess it wasn't a good baseline...

[asmodean]

Sure looks like they suck. Oh well, Intel has always sucked at math. And prolly always will, eh?

[castle lager fan]

I think it's unfair to say Intel suck at Math.

The Pentium 4 will be pretty good at Math when it's SSE2 extensions are used. And then there will the new BIG BOY from Intel in workstation the Itanium. I saw some of it's results running Unix and it's pretty good, especially in the floating point area.

By the way can I maybe get the source code for the program's math as I would like to see what it will do on some RISC machines.

Just a little note. I noticed the program updates the screen on the complex math. Maybe look at the routines to ensure that there are no screen updates in threads during processing as this will influence results. The process might be sleeping for some time while waiting for the screen to update. Just a thought, great program though. I like the fact that there is something where we test the performance of the system a bit not just Hz. It is a better reflection of what people can do with machines than just processor speed. As I have not seen anybody with a machine that do more than 1.6 GHz, I must assume that the scores weren't good or the machines weren't stable enough to run the benchmark.

By the way for some reason my machine sucks at 3d2001 from madonion. I only get a score of 2016 at best. I tried loading the AMD 761 5.22 drivers, but it does not want to load on my Mobo (FIC AD11). Has anybody got any ideas?

Come on somebody beat my score of below 34 secs, maybe it can inspire us to break 30 secs?!? ;-)

[Hoot]

castle lager fan (May 29, 2001 10:28 p.m.):
Hi Hoot,

How do you keep your CPU so cool? I keep going to 45-46 C on mine with the swiftech 462 and Delta fan.

Are you going to try for more?

By the way my motherboard is a FIC AD11.

Where's all the big guns with 1.6 GHz plus? I would like to see how fast those machines go, as I'm unable to break 1.6 GHz yet.

PS : Hoot I'm back at 1.466 GHz as my CPU get's too HOT!! (40C Idle 46C Load Prime 95). My case is also a hot house.

Well, if you read my article on Cohesive Air Cooling, that's how I keep it so cool.
http://www.overclockers.com/tips461/index.asp

That's as far as I'm going to get this 1.0G/266 to go. It sure was fun getting there. Makes me wonder what I could accomplish if I had the luxury of a 1.33G. ;D

Hoot

[asmodean]

castle lager fan (May 30, 2001 12:18 p.m.):
I think it's unfair to say Intel suck at Math.

The Pentium 4 will be pretty good at Math when it's SSE2 extensions are used. And then there will the new BIG BOY from Intel in workstation the Itanium. I saw some of it's results running Unix and it's pretty good, especially in the floating point area.


It's not unfair, they have a bigger share of the market... :)

We shall see about Pentium4... compared to Athlon4!
And Itanium? It's not compatible with ANYTHING. AMD's Hammer family will be. Itanium's not for your average overclocker. Too expensive and McKinley is lurking around the corner.
I'm not trying to start a flame war here, it's just that Intel's CPU's suck at the moment, IMHO. They're good for stability, I guess. And sticking to the past.
Oh well, so's AMD.
Itanium should be named 486, as there hasn't been any MAJOR changes in the architechture since 386. The bit thing, you know...

[FNG]

don't dream too much about the wonder of the 1.33. remember, here is where it won the 'blowtorch on the head of a pin award'. i'm starting to regret not getting a 1.2 or a 1.0 with a better mobo. I can't oc without hitting astronomical temps and can't change the mults on my cpu either cos i have the a7m266. castle lager is tempting me to take this higher but i already broke one axia y, don't need to break this one too. btw, i have the swiftech mc-370 with a 26cfm fan, as2 and about 5 case fans, one PSU fan, blorb on the vid card. idle at 41C, load at 49C avg 43C vcore=1.82-1.89. my mm score was 40.08 seconds. guess i better read hoot's article...

[Bhunta]

hey castle lager fan

my program displays the results with a simple 'DoEvents' statement
It takes virtually to time to accomplish, and keeps the user from falling aleep while watching the benchmark :)
also the timer stops in between the results to ensure no time is lost
take a look at the source its
here (http://www.31337.s5.com)

[castle lager fan]

asmodean(r) (May 30, 2001 01:57 p.m.):

It's not unfair, they have a bigger share of the market... :)

We shall see about Pentium4... compared to Athlon4!
And Itanium? It's not compatible with ANYTHING. AMD's Hammer family will be. Itanium's not for your average overclocker. Too expensive and McKinley is lurking around the corner.
I'm not trying to start a flame war here, it's just that Intel's CPU's suck at the moment, IMHO. They're good for stability, I guess. And sticking to the past.
Oh well, so's AMD.
Itanium should be named 486, as there hasn't been any MAJOR changes in the architechture since 386. The bit thing, you know...

Well, maybe you would have noticed I'm using a T-Bird. So I'm no big Intel supporter, but I also believe they can't just be dismissed. As for Itanium, it's a 64 bit EPIC processor with lots of room in the architecture to grow towards. It is not intended for commercial use yet, so there is very little software for it. There will also be very little. Just for completeness sake see the official spec website @ www.spec.org . It shows a slightly different picture to what can be done with a P4. Pitty Intel did not invite any real software vendors to the party. So for the time AMD rules and I hope they can keep their good effort up!! It is wonderful to see some real competition in this space after such a long time Intal and MS ruled. Enough said!!

Lets try to break 30 secs!!!

[Flanagun]

Last day of school was looking at the thread in comp tech so i tried out the iPaq im using not a horrible score it is 144.758

[asmodean]

castle lager fan: Did a quick look up @ www.spec.org, didn't find anything. Oh well, it doesn't matter. P4 outright sucks at the moment, maybe it'll get better in the future, with better support.

Just tried to run CPUMathMark under linux, wine didn't like my w2k. Maybe I'll try some configuration stuff on it...

[Yozza]

Here's my score in WinME. I got a much lower score in WinXP for some reason - around 47 secs :(

[img="http://homepage.ntlworld.com/yaoyaoliu/math4.GIF"]

Yo\

[Bhunta]

hey
the marks are worse in windowsXP
this is something I am counting microsoft will fix, as I think it has something to do with the visual basic runtimes not interacting with the hardware properly.. This should most definatley be fixed in it's final release, and the scores will be more like the ones in windoes9x, and NT.

[Cez]

I got 38 Seconds!
Will post the jpg tonight when i get home from work!
And this was on a AXIA 1000 o/c to 1333@10*133 (stable speed)

[castle lager fan]

The spec.org I was referring too is located at
http://www.spec.org/osg/cpu2000/results/res2001q1/ (http://) for the AMD and
http://www.spec.org/osg/cpu2000/results/res2001q2/ (http://) for the P4. It shows that the P4 is capable in raw format. There is just nothing that can use it apart from some specialist programs like Flask MPEG4 encoding.
Do we have some actual scores with a P4?
Bhunda I had a quick look at the source code. With the PI calculation it does seem to update the screen some 275 times to the textbox with a DoEvents as well. This will have a bad effect on the scores for this calculation. I will have to dust the rust of my finger and start some real coding again for the first time in years. Would you be interested in C functions in a DLL for the Math calculations? This way we can relate it more directly to other OS and systems, like RISC and Linux/Unix.

I'm interested to see if C wil be much faster than VB with this math. Does anyone have some good assembly tools and references for AMD and Intel chips? This way we can even put some assembly into it, which will make it fly.

Bhunta can we send you the source to control?
Any takers for this new challenge?

I'm still waiting for the real fast machines. A 1.8 GHz Athlon should clear 30 secs. I'm unable to get to 1.6 GHz at the moment, need some more work to be done.

[Bhunta]

hey..
C would most likely give better results than VB, as C can interact direcly to the OS, not indirectly through vb runtimes... therefore giving better results.
I don't know very much C though.. a little bit.. but it's completely different from Vb, its more like javascript or something.
If someone compiled a .dll to be accessed in visual basic that did the math functions that would be great :)
It does work pretty much on scale the way it is though.. this would be a big change and would require a lot of time. Right now im working on some sort of database that you can upload your score to, and the server keeps track of the top scores.. and then possibly some error checks to see if your cpu is generating errors.

[wildone]

here is my 900tb ,it sure beats the old k62

[!-=sky=-!]

duron 750@990(7.5 * 131)

works at 1000 but won't complete the test--got stuck at the 9^1500 part

[asmodean]

castle lager fan (Jun 01, 2001 10:02 a.m.):
Do we have some actual scores with a P4?
Bhunda I had a quick look at the source code. With the PI calculation it does seem to update the screen some 275 times to the textbox with a DoEvents as well. This will have a bad effect on the scores for this calculation. I will have to dust the rust of my finger and start some real coding again for the first time in years. Would you be interested in C functions in a DLL for the Math calculations? This way we can relate it more directly to other OS and systems, like RISC and Linux/Unix.

I'm interested to see if C wil be much faster than VB with this math. Does anyone have some good assembly tools and references for AMD and Intel chips? This way we can even put some assembly into it, which will make it fly.

Bhunta can we send you the source to control?
Any takers for this new challenge?

I'm still waiting for the real fast machines. A 1.8 GHz Athlon should clear 30 secs. I'm unable to get to 1.6 GHz at the moment, need some more work to be done.

I can work on the linux side of the C code. That'll give me another programming project, beside the linux version of InstallShield.

I can dig up a PDF, which contains information on optimizing code for Athlon. It's written by AMD, I believe.
Found it: http://www.amd.com/products/cpg/athlon/techdocs/pdf/22007.pdf
It's 3162k.

[castle lager fan]

[Oc]asmodean (Jun 02, 2001 02:37 a.m.):
I can work on the linux side of the C code. That'll give me another programming project, beside the linux version of InstallShield.

I can dig up a PDF, which contains information on optimizing code for Athlon. It's written by AMD, I believe.
Found it: http://www.amd.com/products/cpg/athlon/techdocs/pdf/22007.pdf
It's 3162k.

Thanx for the link. I will looak at it. I just got my PC up again tonight s I will start working on it this week.

[Riekk]

"Error starting program"

"A required .DLL file, MSVBSM60.DLL, was not found."

[asmodean]

Riekk (Jun 04, 2001 01:02 a.m.):
"Error starting program"

"A required .DLL file, MSVBSM60.DLL, was not found."

You are missing the VB6 runtimes...

[Bhunta]

get VB 6 runtimes here (http://www.tweakfiles.com/cgi-bin/dl-ftp.cgi?ftp.zdnet.com/threedee/tweakfiles/misc/vbrun60.exe)

[GERRY136]

sigh.

[Fjeld]

Hi Bhunta!
Great stuff, that Mathmark-thing! And of course, I gave it a try. And sendt my AVIA Y 1133 T-bird into a screaming 1532mhz. Watercooled of course, but at a very premature stage. My wateblock connected to a 2$ pump, submerged in a 1,5ltr coke bottle(!). (Yes, I´m watching it all the time!!)
My result with this rig: 35,212sec´s! Not bad,eh? :) I don´t remember my bios-settings, but my memory stucks@140mhz! So I guess I had the multiplier for 11x or so. I don´t remember. Sorry to say, I have no screenshot so far, just because I don´t know how to get one!
Please tell me, how do I get a screen-shot??

Regards
Fjeld@Norway

[asmodean]

PaintShop Pro can take screenshots, but I think using it's ability to capture just a window is more useful here, no?

[TranceBear]

got some new cooling thanks to Hoot!! I am over the 1.5ghz mark, what a good feeling!!

[Fjeld]

Hey!!
At last, got a screenshot, but not the very best :( But still, a reasonably good result vith my AVIA Y T-bird 1133!?

[castle lager fan]

Bhunta,

I just amanged to break 1.6 Ghz so here's my last score. Still not able to break 33.3s. I wanted to go higher but my mobo and memory was too unstable. I also cant run Prime 95 at 1.6 as it gives errors, but your program works fine. See another posting i'm going to put in.

[castle lager fan]

OOPS

[!-=sky=-!]

did another one in win 2k

[Kramerica]

here is mine...would ddr memory lower the time at all?

[Bhunta]

CPUMathMark 3.0 will be coming out soon
I need some baseline results and I'm interested if anyone is willing to give me theirs
just post your email
thanks

[TranceBear]

BamBam@Luv2Rave.com , no problem

[Fjeld]

Hm, I may come down inside the top then with this?
KT7A RAID, 1133T-bird AVIA "Y"@1577, 256Mb MOSEL-ram, 30gb IBM ATA100, CardExpert GF2Pro400 64Mb, SB Live Player5.1
Watercooled :)

[Fjeld]

Bhunta (May 26, 2001 10:34 p.m.):
Im looking to see what other people get in high end AMD systems
You may have seen the program at tweakfiles lately, but I've been getting little or no high end amd score feedback. (I am one of the two authors of it)
you can download it here (http://www.tweakfiles.com/cgi-bin/dl-ftp.cgi?ftp.zdnet.com/threedee/tweakfiles/benchmark/CPUmathMark.exe)
my 112x8.5 (952MHz) Duron gets 57.3 seconds


Hm, I may come down inside the top then with this?
KT7A RAID, 1133T-bird AVIA "Y"@1577, 256Mb MOSEL-ram, 30gb IBM ATA100, CardExpert GF2Pro400 64Mb, SB Live Player5.1
Watercooled

[GERRY136]

damn,u guys rok!!!!!!!


ill be there

[Hoot]

Wait until you see the next version, currently out in beta. The code and calculations are much more optimized, as far as I can tell. Runs the test in about 14 seconds now.

Hoot

[TranceBear]

Can't wait!!

[erickkk]

My score :

[erickkk]

[img="http://erickkk.multimania.com/Overclocking/MathMark.JPG"]

Nice ;D

[Sharp]

Just a suggestion for your benchmark program. Have it create an encrypted file with the results. That'll keep the average user from boasting outrageous scores, you can ask them for the file along with the screenshot. Anyone can easily doctor a screenshot but few could/would crack even a simple encryption scheme. Also if you had a graphical silkscreen behind the text boxes you could prevent the average paint owner from doctoring it up... I can't wait to get my 1.2 MP so I can test it with this benchmark [it is supposed to have much more powerful + faster math.] I don't know if it has to be compiled optimized for the core though ;-( It can return the results for 4 single precision math operations per clock cycle :-) but I guess your program probably uses doubles

[Hoot]

erickkk (Jul 18, 2001 10:30 p.m.):
My score :


My hat's off to you sir. That is really honkin' !!!

Hoot

[Fjeld]

erickkk (Jul 18, 2001 10:30 p.m.):
My score :


GOSH!! I just lost my breath!!

What kind of cooling? Pelt´s and stuff? Nitrogen? Just curious!

[Bhunta]

Everyone is giving good suggestions and I am doing my best to cram everything in I can until the release date which is this coming monday. The first place to get it at will be
31337 (http://www.31337.s5.com)
then tweakfiles
I want to cover a range from ~1.7GHz - ~300MHz for benchmark comparisons
2 tbird, 2 duron, 2 PIII, 1 PII, 1 Celeron, 1 K62 and whatever I have left to throw in for one
(maybe. cyrix wait.. ha)
this sound good?[

[erickkk]

Fjeld (Jul 19, 2001 10:53 a.m.):
erickkk (Jul 18, 2001 10:30 p.m.):
My score :


GOSH!! I just lost my breath!!

What kind of cooling? Pelt´s and stuff? Nitrogen? Just curious!

Thanks, you can see more on :
http://forums.overclockers.ws/forums/UltraBoard.cgi?action=Read&BID=2&TID=24691&SID=113704

[el]

I knew this guy would find this thread. Nice score erikk :¬)

[ChapFou]

[img="http://macsquare.free.fr/photos/bench3.jpg"]

[Bhunta]

thanks for all the support everyone
if only I got this level of support in the intel section...
:)

[TranceBear]

Bhunta (Jul 19, 2001 10:27 p.m.):
thanks for all the support everyone
if only I got this level of support in the intel section...
:)

They are too busy figuring out why AMD chips are so much faster..hehe

[Bhunta]

hmmm never thought of that
It must be embarassing when you have a p4 1.7Ghz to have people with 1.3GHz athlons and 1.2 GHz althon MP's keeping right up with you
I heard the new tualiton PIII's will outperform the p4's.. by a lot.
now THAT would be funny if intel dropped p4's altogethar...
Can someone say biggest market flop ever? :)

[JRbob]

I cant get the program to run keeps telling me DLL file MSVBVM60.DLL
is missing.Can someone help with this.

[Jetset]

I do the benchmark in 52 seconds with a 1.33ghz tbird with 256mb ddr. Anyone know why I am scoring so high?

[Bhunta]

JRBOB: you need da visual basic runtimes

[Bhunta]

We were finally able to release CPUMathMark 3.0 last night check out31337 (http://www.31337.s5.com) for download / info, and tweakfiles once it's up.
Heres my screenie

[TranceBear]

Once again, I get a better score with a lower mhz, why is that? well, anyway here is the new score.

[Bhunta]

How hot does it run when it's at higher MHz? Electric currents have less resistance in cooler temperatures, so is there is a big difference in temp this could be it (however ~5C shouldn't matter) Free RAM will also matter a bit in the score, bios settings, how long your puter has been on..
just throwing out ideas here, as this confuses me too

[rezo]

here is my best score...

[GRiMMi]

Ummm my 800 duron did this in 28 seconds on the newest math mark version... looks a bit weird opposed to the other scores.. ?

Gonna post a pic after

[Bhunta]

your using version 3.0 and most of the scores in this form are from 2.1
3.0 runs in approx 1/2 the time of verson 2.1 that'd be why :)

[dozier768]

rezo (Jul 27, 2001 03:00 p.m.):
here is my best score...


a duron @1.3 + ghz :-O~ drool drool

[Newbie_Doo]

How about a poor overworked Celeron...(I know, not an AMD CPU, but I was here and all...)

Celeron2-600@945 watercooled:

[Newbie_Doo]

Here is the score under 3.0. Quite a difference from V2.1.

[FRANK]

AXIAR-Y 1G @ 1.42
This was done with v2.1. Whenever I try to download v3.0 I get a prompt for a password.

[ZX-Tex]

OK here is my screen shot. I have run this computer with watercooling but am experimenting with modified aircooling. For this test I ran at 142x10. This is a T-bird 1200.

BTW - I am new to the site, this forum is supermegahellastylin.

[TranceBear]

For those guys who are using ver 2.1, your scores will be double what the new version is doing. Scroll up to get the website to the new version.

[ZX-Tex]

TranceBear (Jul 28, 2001 12:58 p.m.):
For those guys who are using ver 2.1, your scores will be double what the new version is doing. Scroll up to get the website to the new version.

I tried running version 3 twice and the program crashed right after the Pi iteration test. I am running Win2K service pack 2. Anyone else seen this?

[erickkk]

V3.0 :

http://erickkk.multimania.com/Overclocking/CPUMathMark3.exe

[erickkk]

[img="http://erickkk.multimania.com/Overclocking/1732cpumath.JPG"]


[img="http://erickkk.multimania.com/Overclocking/1718cpumark156.JPG"]

;D ;D

[castle lager fan]

Good stuff Erickkk!
Here's my score!

For those of you interested. V3 uses C routines instead of VB for the math. As C is about 30-40 times faster than VB, the iterations had to be increased to still give it a meaningful benchmark result.

Here's the list of how many times the iterations had been increased:
Addition,subtract,multiply,divide 16.6 times
Prime 3 times
Pi 9 times
Power 300000 times (as VB had to use strings for big number)

Interesting!

[erickkk]

night score for me ;D
@1740 10*174 16-8-8-2-2-2-2

[img="http://erickkk.multimania.com/Overclocking/1740et159cpumark.JPG"]

[img="http://erickkk.multimania.com/Overclocking/1740cpumath.JPG"]

[castle lager fan]

Looks like you outpaced me!!
Here's my last try at it.

[Bhunta]

Those are some nice results
put my duron to shame

[Eil Atan]

man my scores are aweful compared to some of the other ones on here... I think I need to actually start overclocking my chip, but these danged temps! ahhhhh! anyway, nice scores all of you people in the below 36 second range, i am indeed envious...

[Bhunta]

Your running v. 2.1 get version 3.0 here. (http://www.tweakfiles.com/benchmark/cpumathmark.html)
scores run in about 1/2 the time :)

[el]

great for burn in too. I only scored 15.875 with my 1.2 at 1.4 truly sad :(

After 100cycles my temp under load reached 50C!!!!

[Bhunta]

what's your temperature get with SETI or other benchmarking utils?
I never really actaully tested it temp wise...

[Sharp]

why in the @!%$ does it require winsock?? sounds very highly fishy...

[TranceBear]

Originally posted by Sharp
why in the @!%$ does it require winsock?? sounds very highly fishy...


You gotta be kidding? , Bhunta has been a valuable help to this forum for alot of things. The biggest is giving us this great new CPU benchmarking software that he did on his own time, you think he got paid for all his time? Go flame elsewhere!!

[TranceBear]

test with new pic

[Fjeld]

Hi Bhunta!
I´m a bit dissapointed, not finding my score in your Hall Of Fame?
Ok, i ran my test in a low res. mode, but still I reached a pretty high score :) 34,29! I have sent you my screenshot earlier, so please have a look in your earlier mail! I´m not trying to be rude here, but just fair!
So if you please....
Regards!
Fjeld

[TranceBear]

I think he may be doing a newer list for the new version 3.0, give hime some time. He has always responded back to me quickly before this new software came out.

[Sharp]

Originally posted by TranceBear
You gotta be kidding? , Bhunta has been a valuable help to this forum for alot of things. The biggest is giving us this great new CPU benchmarking software that he did on his own time, you think he got paid for all his time? Go flame elsewhere!!

I'm not kidding, and I don't doubt he has been a valuable help. I too am a programmer and I have never been paid to do it. I help around here whenever I have the chance but as you can see I haven't been here too long. I guess I phrased the question wrong because it was not a flame. I don't do flames. Anyways it is very odd that the winsock files should be needed. These files are used for socket communications [TCP protocol mostly.] Unless it sends scores to the server or something it doesn't need to be there. If I had access to the source code I could tell you exactly what it is used for [quickly], but the source seems to have disappeared. I guess I'll just have to sniff the traffic when I have time. I happen to be actively security conscious and credited with several exploit discoveries. I have made no accusations and I am just curious why it would be there, I am sure no harm is meant by its requirement and it is safe to use... just curious

[Sharp]

I have sniffed the traffic while the program is running and everything is ok. The winsock files are needed to send the results in to the server [I didn't see the option before.] On a side note the feature didn't work... probably got shot down by my firewall. Result submission by www server and a simple cgi script could easily bypass that problem though.

[Jon]

Nifty little program. I like it.

[N3D2000]

Duron 800 @ 840

64.662

I wonder what it would be at 1050?

[Tomas]

After getting my 1.4g and oc too 11.5(133) 1547mhz, dunno why i get nearly 135fsb when i put it at 133....

But got a score of 35,17seconds. is that good?

[Bhunta]

Sharp:
It uses winsock controls because of the submit results feature. However this feature did not turn out the way I thought it would as it requires more bandwidth then I thought it would.


Fjeld:
Sorry about the hall of fame :)
I have been busy this summer with many things and have not had time to make one for v.3.0

Other good news... I recently recieved an email from a person that works at intel Francois Peidnoel and he is going to put in optimizations in CPUMathMark for the p4 and PIII. I also recieved an email from Loyd Case from ziffdavis.com and he might be doing an article involving CPUMathMark, but I need not say more as I don't want to ruin his article :)

[TranceBear]

Originally posted by Tomas
After getting my 1.4g and oc too 11.5(133) 1547mhz, dunno why i get nearly 135fsb when i put it at 133....

But got a score of 35,17seconds. is that good?


That is still using the older version of this software, version 3.0 will get your scores cut in half. It changes the way the calculations are made. This screen shot is a bit old, but if gives ya the idea.

[N3D2000]

59.511 Duron 800 @ 920 8x115

WOW..... my comp is kinda hot now.....

[N3D2000]

25.04 w/ version 3.0

[TranceBear]

*edit* with better pic

[Bhunta]

nice tracebear
This will be interesting to see once the optimizations are done :)

[!-=sky=-!]

version 3 is a lot faster....

[erickkk]

http://erickkk.multimania.com/Overclocking/cpumathmark1790.JPG

[Liquidsky]

I finished the test in 25.34 seconds on my Duron 900 (9x100) with 320 MB of RAM. My Celeron 1064 took 31.14 seconds.

[Bender]

Here is my athlon at 1602.

[TranceBear]

Glad to see ya got it working Bender

[Punk]

i got 15.612 in mathmark with my AMD thunderbird 1333 running @ 1433MHz.

here is a screenshot:

[PanteraGSTK]

My 1ghz @ 1.53