cpu variation error

[|RickY|]

this ****** error is verry anoing :(

i made 26k with p4 3.2 and... this error

and now i made 27k with asus k8v + a64 and this §§§§ing error again...

i start win at 240 fsb and I use clockgen before oppening 3dmark and set it to 258 when i open 3dmark i see the cpu detected as 2580 mhz....

I have clock down to 254 to run nature with better clocks on 9800pro (agp is too high)

and at the end of 3dmark the cpu is detected as 2580 too

but.... this error again and again

http://www.xtremesystems.org/forums/attachment.php?s=&postid=266498

i'v 2 runs of 3dmark and this error are always there

any way to prevent this anoying error?

regards
:toast:

[Stang_Man]

i know, i hate this.

open up 3dmark2001, and check your system details
if anything is incorrect, close then open it up again

hopefully you'll score 27k again

[|RickY|]

i'v cheked and was right :S

i can't publish good scores... :mad:

going to try to dont use clockgen... but it was a usefull program :S

tanx anyway :S

[vlad]

I've had the same problem. it seem to happen when i first test a new OC, but once its ran at the new speed for a bit i can publish.
my 2 top scores right now are unpublishable. that BLOWS!

[Edward2]

I guess I have either been lucky or don't push my computer hard enough. I have never had this problem with either of my rigs (IC7-G and NF7). <knocks on wood repeatedly> I don't use clockgen while benchmarking however, might that be part of the problem?

[Garrett]

I had that problem too... turned out I had thermal throttling enabled in my BIOS...
Disabled it and I could publish my scores again :)

[Marci]

More or less any software based FSB Adjustment tool causes this error in my experience... you can't seem to be able to change the FSB of a system once it's booted without this error occuring. Clock changes must be applied on post thru the bios or it'll b0rk yer score...

[Kunaak]

can you try this.

run the entire test at 2540.
you changing the CPU speed is whats giving the CPU variation error, or your CPU isn't stable enough for the test.

PS.

Would you mind editing that title of your test in the pick?

;)

[|RickY|]

sorry about the **** :rolleyes:

tanx for all i'v done very test using setfsb and variating the fsb, and i could submit...

and i'v done some tests without variating fsb and it gives cpu vriation... anyway there is my last 3dmark score (fsb variation:(258~260))

17104 (http://service.futuremark.com/compare?2k1=7211730)

need a board with agp lock to get more marks ;)

tanx for all
:toast:

[Kunaak]

your scores off by about 10,000 points ;)

[mkleingeld]

This is a kinda old topic but I want to express my annoyance as well.
I scored 23556. The number 1 score with a R9700pro. Finally passed JC's old score. And guess what. Same iritating error and I can't publish :(
I too use clockgen because of the agressive GAT settings I can't post at full speed.
:(

[Marci]

It's incorporated to catch people who were running different FSBs for every test... usually considered cheatin' from what i recall... all valid benches are done at a constant FSB thru-out... don't think Macci, Opp, JC, Icee or any of the other guys move the FSB between tests for this reason...

Not sure how using a TurboPLL affects this.... Macci'd be able to answer that question... but in the days when Macci used to use one 3DMark didn't have anticheat patches in etc, and whether it was cheating or not was a grey area... nowadays however it's a different story...

[mkleingeld]

I just tested it.
I think the error comes when you first open 3dmark and then set the fsb.
I just rebooted with 150 fsb, set it in windows to 258 (no need to stress for a test) then opened 3dmark and I could publish without problem. Going for a number 1 run this evening (no 1 in the class) :)

[StormPC]

Originally posted by Marci
It's incorporated to catch people who were running different FSBs for every test... usually considered cheatin' from what i recall... all valid benches are done at a constant FSB thru-out... don't think Macci, Opp, JC, Icee or any of the other guys move the FSB between tests for this reason...

Not sure how using a TurboPLL affects this.... Macci'd be able to answer that question... but in the days when Macci used to use one 3DMark didn't have anticheat patches in etc, and whether it was cheating or not was a grey area... nowadays however it's a different story...

JC changed his for car high, and I'm sure others do it too.

No, it's not cheating any more than disabling services or hardware in Windows or the BIOS is cheating.;)

[MrIcee]

I don't in fact change my FSB at all when running the bench, as Marci stated. I'm more apt to reduce a card clock to prevent freezing in a certain test or tests.:)

There is a way that can help prevent the CPU variation problem, and I ran across it awhile ago. It is usually caused by a slightly unstable overclock. It may bench just fine, but the CPU is doing a little wobble game in its Mhz.

When you are at your top CPU benching Mhz, open CPU-Z. A stable overclock will not flucuate on the opening screen. An overclock that is unstable will show the FSB speed flucuating up and down by a few Mhz. When it's doing that....you will not be able to publish scores because 3Dmark senses it as a CPU speed change outside it's cheating parameters.

This problem manifested itself late last year...when a few of us got NF7-S's in our hands. That problem was a major roadblock then and is still a pain in the arse now. The best way around it if you suffer the non-publishing symptom is to back down the FSB 1Mhz at a time until it will publish.:)

Randi:D

[StormPC]

Yeah, I had that problem with my A7N8X2.0 board when I was pushing it on the FSB.

Thanks Captain!:)

I didn't mean that everyone does it, I just said it's not that uncommon and could not be considered cheating.;)

[JCviggen]

In my experience the CPU variation error is not really related to wether CPU speed stays constant through the benches...more like when you open 3Dmark the speed is recognized incorrectly. Pushing the OC too far can give this effect on some systems but on some it doesnt so...its hard to say.
I sometimes run car high 50 MHz higher than the other tests and it doesnt give me any crap at all about CPU variation. I did lose my 29K run to that but that was my fault by changing the multi in windows and 3DM didnt see it :(

[mkleingeld]

offtopic, I just missed out on your r9700pro #1 slot with a measly 4 points JC!
But had a bad car low some 20fps below what my rid can do.
Sunday night it will freeze here :)

Ontopic, yea, now I learn first to check if 3dmark correctly recognizes the cpu speed before even attemping a run.

[macci]

I have clock down to 254 to run nature
Thats the reason for the 'cpu variation error'. The error seems to appear only if you change clocks when mark2001 is running. I had it once too on P4 when I lowered CPU clocks for nature using clockgen.

If you change the multiplier or clocks before you start mark2001 no such error will appear (excluding NF2 errors..).
For instance on A64 I always boot up at 210x10, then adjust it to 9X multi with clockgen and then adjust the FSB in the 320+ range and then fire up 3dmark - haven't got a single cpu variation error by doing this :)

Ricky, you need to let the K8V and VIA go. get yourself a nice Gigabyte NF3 and you'll get much much better results ;)

[|RickY|]

tankyou ppl

i'm starting to understand the cpu variation error :D

but sometimes it detects sometimes not :S

anyway now i dont have cpu variatio errors :)

macci: of course i want a nf3 board but... in Portugal I only know 1 more person using a64... its hard to sell my via board, i'm waiting for nf3 250, buying a gigabyte now is losting very Euros ;)

waiting to get a good 9800xt and do some damage with the via boad :) now 27.8k, 28k easy with a good xt

Waiting to see u with 30k on a64 :D

:toast: