best super pi times?

Printable View

Show 100 post(s) from this thread on one page

11-07-2004, 01:27 PM
Zeus

Guys,

I have a question about SuperPi.
If i run SPi 1m, 2m or 4m CPU utilization is 100% throughout the entire run but as soon as i try 8m or higher, CPU utilization fluctuate between 0 and 100%.
Also, 32m took 44minutes to complete at 10x235, not good at all i think.

I have no programs running in the background.

WTF is causing this? :confused:
Anyone else with this problem?

http://www.dc3k.com/hosted/getimg/CPU_util.jpg

Look at the CPU utilization statistic diagram. :(
11-07-2004, 01:38 PM
Learn

Quote:

Originally Posted by Zeus

Guys,

I have a question about SuperPi.
If i run SPi 1m, 2m or 4m CPU utilization is 100% throughout the entire run but as soon as i try 8m or higher, CPU utilization fluctuate between 0 and 100%.
Also, 32m took 44minutes to complete at 10x235, not good at all i think.

I have no programs running in the background.

WTF is causing this? :confused:
Anyone else with this problem?

http://www.dc3k.com/hosted/getimg/CPU_util.jpg

Look at the CPU utilization statistic diagram. :(

it's windows 2000? use windows xp or windows 2003 server :)
11-07-2004, 01:41 PM
Zeus

Is win2K the culprit??
11-07-2004, 01:48 PM
bachus_anonym

i'm not sure but it might be caused by some sort of instability...
what sometimes happens (if you look at processes window) is a process called drwtsn.exe kicks in stealing cpu clocks... i'm not sure what causes debugger to kick in....
i got that sometimes when i push FSB to high on my Infinity...
11-07-2004, 01:49 PM
Learn

Quote:

Originally Posted by Zeus

Is win2K the culprit??

yes, it's bad with super pi > 1M
11-07-2004, 01:54 PM
Zeus

I can imagine that but 2350MHz isn't that high i'd say, it's prime stable at that speed.
If i remember right i have this at default speed as well, have to try again to be sure.
Thanks for your input anyway. ;)
11-07-2004, 05:11 PM
cpulloverclock

2 f**ked and faked scores :rolleyes: :mad:

19 & 21s...
http://www16.big.or.jp/~bunnywk/cgi-bin/superpi.cgi
11-08-2004, 08:19 AM
Tom Holck

Quote:

Originally Posted by misteroadster

I've done 26sec @ 30..Mhz with my FX55 , opp did 24 sec at 34..Mhz , Tom 22 at 36..Mhz ??
I deduct there's 2 sec missing between 3000 and 3400Mhz. Strange. :confused:

I'm at 3730 64 bit OS misteroadster.

Depending of OS and tweaks and Ram-speed:

25 sec 3120 Mhz (64 OS) 940 Mobo
24 sec 3350 Mhz (32 OS) 940
23 sec 3520 Mhz (64 OS) 939
22 sec 3730 Mhz (64 OS) 939
21 sec ? my gess 3950

Btw nice FSB speed you got, my max is 313 then Ram is 3-4-4-8 :toast:

Quote:

Originally Posted by cpulloverclock

2 f**ked and faked scores :rolleyes: :mad:

19 & 21s...
http://www16.big.or.jp/~bunnywk/cgi-bin/superpi.cgi

Yes, the one is now removed, but still 21/3825/975/244 must be fake at least at least 100 mhz to slow, and slow FSB :(
11-08-2004, 11:09 AM
misteroadster

Okay tom, are you sure 2003 64 bits is faster than 2003 32bits?
25 sec at 3120 on 940 is very impressive !!!!
Can you try to analyse the progression of the 939 and the same 64 or 32bits OS when you ll got the time ?
As you could see on my screenshots :
the best for 1sec gain :2547-2619=80mhz with same settings, just FSB increased.
After that 1sec again between : 2619-2717 =100mhz same setting, just FSB increased.
And 2717-2847=130mhz same settings but FSB-20Mhz.

I've done 26sec at 3060 but i can do better now , 2847-3060 =213 mhz/2sec.
I understand your FX 939 is not better than your 940 setup , normally it's 1 sec better easy.
Give a try with your ECC sticks , i think it'll be interresting.

Tom turn on my light please , and of course congrats, if i was nearest from you , i'll lend you my BH-5 and my X800 :toast:
11-08-2004, 11:16 AM
cpulloverclock

Tom u dont want to test a dothan + aopen with your cascade? 100% compatibility
11-08-2004, 01:00 PM
Tom Holck

Quote:

Originally Posted by misteroadster

Okay tom, are you sure 2003 64 bits is faster than 2003 32bits?
25 sec at 3120 on 940 is very impressive !!!!
Can you try to analyse the progression of the 939 and the same 64 or 32bits OS when you ll got the time ?
As you could see on my screenshots :
the best for 1sec gain :2547-2619=80mhz with same settings, just FSB increased.
After that 1sec again between : 2619-2717 =100mhz same setting, just FSB increased.
And 2717-2847=130mhz same settings but FSB-20Mhz.

I've done 26sec at 3060 but i can do better now , 2847-3060 =213 mhz/2sec.
I understand your FX 939 is not better than your 940 setup , normally it's 1 sec better easy.
Give a try with your ECC sticks , i think it'll be interresting.

Tom turn on my light please , and of course congrats, if i was nearest from you , i'll lend you my BH-5 and my X800 :toast:

I'm sure 64 is faster in Pifast and SuperPi with the same clock, but more heat, that can affeckt the max OC. And then it's not faster.
I'll compair 32/64 on 939 later. But have done that on the 940, difference in performance 32/64 was aprox 60hz in arround 31-3200. The force of the 940 was stability and stability... High CoreVolt (2.112 stabil) The ECC was stabil at 262 fastest settings, thats close to 939 performance.

Thanks for the offer. I'll get my 2. x800xt in a few dayes. But it take a lot to insulate and make it work on the cascades. Huge kondensation.
Another thing the MSI Neo2 is very small for my coolingheads. The AGP is close to the cooling head without insulation :eek:

Quote:

Originally Posted by cpulloverclock

Tom u dont want to test a dothan + aopen with your cascade? 100% compatibility

Tel me about it!?
11-08-2004, 01:23 PM
aoc007

Dothan on a cascade? Now that I would love to see :slobber:
11-08-2004, 02:12 PM
cpulloverclock

Quote:

Originally Posted by Tom Holck

I'm sure 64 is faster in Pifast and SuperPi with the same clock, but more heat, that can affeckt the max OC. And then it's not faster.
I'll compair 32/64 on 939 later. But have done that on the 940, difference in performance 32/64 was aprox 60hz in arround 31-3200. The force of the 940 was stability and stability... High CoreVolt (2.112 stabil) The ECC was stabil at 262 fastest settings, thats close to 939 performance.

Thanks for the offer. I'll get my 2. x800xt in a few dayes. But it take a lot to insulate and make it work on the cascades. Huge kondensation.
Another thing the MSI Neo2 is very small for my coolingheads. The AGP is close to the cooling head without insulation :eek:

Tel me about it!?

a dothan at 3.8GHz is more powerful than a FX-55 at the same frequency
11-08-2004, 03:11 PM
Tom Holck

Quote:

Originally Posted by cpulloverclock

a dothan at 3.8GHz is more powerful than a FX-55 at the same frequency

OK, when do I test it :banana4:
11-08-2004, 10:31 PM
zakelwe

Excellent Superpi and pifast scores Tom. I particularily like the pifast score :)

Edging slowly towards the magic 30second barrier ( some way to go though ).

Regards

Andy
11-08-2004, 11:39 PM
misteroadster

Quote:

Originally Posted by Tom Holck

I'm sure 64 is faster in Pifast and SuperPi with the same clock, but more heat, that can affeckt the max OC. And then it's not faster.
I'll compair 32/64 on 939 later. But have done that on the 940, difference in performance 32/64 was aprox 60hz in arround 31-3200. The force of the 940 was stability and stability... High CoreVolt (2.112 stabil) The ECC was stabil at 262 fastest settings, thats close to 939 performance.

Thanks for the offer. I'll get my 2. x800xt in a few dayes. But it take a lot to insulate and make it work on the cascades. Huge kondensation.
Another thing the MSI Neo2 is very small for my coolingheads. The AGP is close to the cooling head without insulation :eek:

Tel me about it!?

Thanks tom.
Can you try yours ECC sticks on 939 if it works ?
Do some tests at same timings with ECC- and non ECC ram , and you ll can touch de diff between 939 and 940.
11-09-2004, 03:43 AM
funkflix

http://people.freenet.de/funkflix/pi30.png

I wonder which clocks i would need, to get the 29 sec. ?!
11-09-2004, 03:55 AM
CTKP

Running 1:1 will get you there.
11-09-2004, 04:05 AM
funkflix

Yeah, BH5 @ 310 2.0-2-2-5! :banana:
11-09-2004, 04:12 AM
misteroadster

Quote:

Originally Posted by CTKP

Running 1:1 will get you there.

Yes, probably :toast: if i could have 30 sec at 2674 With winXp you can have 29 sec at 2780 easy with 2K3 server and 278 2/2/2/1 1:1.
http://pageperso.aol.fr/misteroadster/350030sec.JPG
11-09-2004, 04:18 AM
funkflix

How much Volts ~ for 275 2.0-2-2-5? 3,75 + i think!
11-09-2004, 04:21 AM
macci

Quote:

I wonder which clocks i would need, to get the 29 sec. ?!

w/ that RAM divider mine reaches 29sec at 2817MHz (11x256) under W2K.
11-09-2004, 04:25 AM
funkflix

How much Volts on ur 3200+ ?Thats exactly the same settings i have, just
a few MHz more. U tweaked it with A64-Tweaker?

Edit: Ah, ok ~ 1,8V.
11-09-2004, 06:24 AM
funkflix

I cant get my 3200+ stable over 2800MHz, doesnt metter how much Volts.
1,76 was the max.! It runs better with less Volts! Do u think the Heatspreader
ist the Problem? Maybe i should cut him off! Can run SuperPi with 2790 MHz
@ 1,59V! Strange! I can run the BH5 at an lower Multi @ 260 2-2-2-5 1T
with 3,45V, so i think thats not the problem!
11-09-2004, 07:05 PM
cpulloverclock

eheh not bad :cool:

I'm 6th of the P4-36GP, I get 33s at 3705MHz :cry:
http://www.biwa.ne.jp/~yok/36-GP/36-GP-TABLE.htm

31s at 4026MHz with only a fsb of 236.8 :)

http://membres.lycos.fr/cpulloverclock79/30...gif
http://membres.lycos.fr/cpulloverclock79/3600.gif
http://membres.lycos.fr/cpulloverclock79/3597.gif

Show 100 post(s) from this thread on one page