Weird Orthos Error

J

jcll2002

2[H]4U
Joined
Jan 28, 2006
Messages
3,655
I just installed a few yate loons and took out some old OEM ones... Before this, (a couple weeks ago) i could orthos for over 6hrs (no error...just stopped it). Now, I run orthos and i get this error after 4 mins on core 0

"FATAL ERROR: Resulting sum was -6.152755342537605e+017, expected: 168358151421872.8
Hardware failure detected, consult stress.txt file.
Torture Test ran 4 minutes 21 seconds - 1 errors, 0 warnings."

Yeah... What do you guys recommend? In the meantime, I'll re-run memtest and/or occt (both of which were passing for hours...well, oct for 1/2 hr :)))
 
jcll2002 said:
like i said, it passed for 6hrs+ until today...
Click to expand...

6+ hrs doesn't mean much. I could pass 27+ hours and still know how to crash it. I've had it pass 27 hours, then fail. I'd hit retry and it'd fail right away.

how are your temps like? It sounds like you were at the brink of stability possibly before and you disrupted airflow pretty badly, causing some devices to be too hot possibly.
 
before there was hardly any airflow, and now my temps are lower. I was loading CPU @ 43c and chipset 47c before the "mod." Now I am like 40c cpu load and 44 chipset load.

I didnt fail orthos after 6hrs, i just stopped it.

oh yeah and PWMIC was like 57c load. (w/ fan over it. i heard it can go way over that though)

EDIT: memtest just passed 100%...so...yeah im really wondering whats going on

i fail again @ 9mins FATAL ERROR: Resulting sum was 9.903520314283042e+030, expected: 7.51965979878813e+016
 
i think the aboce poster suggesting more ovlatge is probally pretty accurate, especialy if your using on board fan headers, if air lor was crap beforer and you added more/higher demanding fans it iwll cause stability issues if you were close before ( close is a fine line i run 24+ hours prime stable at 1.45 volts 2.7ghz on my 165 at 1.40 i cant even do 2.3 stable) and i cant do 2.8 to save my life so there ya go. try more volatge, if your using motherboard fan headers ditch them for a fan controller. the rounding error from what ive read is 7/10 times voltage related not necessarily heat.
 
ok i guess ill try adding more vcore. BTW its 1.475v in bios and 1.43v in windows. Is that vdroop normal?

I'm trying 1.485 in bios
 
I was failing after 6-10 minutes, until I upped the vCore. Now I'm at an hour and 20 minutes and so far so good.

vCore seems to be the culprit, jut don't go too high.
 
i just dont understand how changing fans...all still undervolted, would affect stability that much?
 
Who knows, your heatsink paste may have settled in, or the extra voltage finally muffed something up, or a tiny bit of static electricity fried something.

I wouldn't take Orthos as the gospel of stability either, or OCCT, or any "stress" tester. They all push the FPU or certain CPU functions nicely to create heat, but until you get into real world apps/games all bets are off.

My X2 ran perfectly stable with Orthos, Prime95, OCCT, etc. even the standard [H] test ran fine for 24+ hours (IOMeter, 3DMark06 looping, triple Prime95 all at the same time). However a few weeks later for giggles I tried the "verify disc" function in Nero when burning a DVD, and it failed, finding all kinds of different sectors (which incidentally make no difference when playing the disc). Also I ran into some weird problems with Quickpar failing unless I had 40% more PAR files than I actually needed. Anyway, at first I thought my DVDRW was bad, no it wasn't. Then I though it was the southbridge overheating, so I upped my intake fan, that made no difference. Then I thought it was my RAM, even though it's running in spec, loosened timings, made no difference. After much troubleshooting (stock settings worked fine), turns out all I needed to do was up the CPU Vcore by 0.025V and all was well.
 
theelectic said:
Who knows, your heatsink paste may have settled in, or the extra voltage finally muffed something up, or a tiny bit of static electricity fried something.

I wouldn't take Orthos as the gospel of stability either, or OCCT, or any "stress" tester. They all push the FPU or certain CPU functions nicely to create heat, but until you get into real world apps/games all bets are off.

My X2 ran perfectly stable with Orthos, Prime95, OCCT, etc. even the standard [H] test ran fine for 24+ hours (IOMeter, 3DMark06 looping, triple Prime95 all at the same time). However a few weeks later for giggles I tried the "verify disc" function in Nero when burning a DVD, and it failed, finding all kinds of different sectors (which incidentally make no difference when playing the disc). Also I ran into some weird problems with Quickpar failing unless I had 40% more PAR files than I actually needed. Anyway, at first I thought my DVDRW was bad, no it wasn't. Then I though it was the southbridge overheating, so I upped my intake fan, that made no difference. Then I thought it was my RAM, even though it's running in spec, loosened timings, made no difference. After much troubleshooting (stock settings worked fine), turns out all I needed to do was up the CPU Vcore by 0.025V and all was well.
Click to expand...

so true. Although in my case, it was easier to tell. For me, running 3.6ghz, I still have a little trouble. Maybe a better psu + o-ring for my water block will help. But as it is right now, I'm back to 3.2 at the moment (although that doesn't break a sweat at all and I might as well be using stock hsf). At 3.6, I passed small fft for almost 25 hours and blend for over 27 hours. Large fft passed for over 14 hours as well before I decided to throw 3dmark into the loop. Basically, with large fft and 3dmark, after an hour or two, it'd die. Hell, large fft as I found out, would die after enough time as well (excess of 25 hours on average). Playing cs would cause it to crash to desktop within a few games (always in between loading maps).

edit: prime/orthos is becoming less indicative of stability nowadays due to the large caches cpu's are coming with. I was informed that small/large/blend all don't necessarily test all spans of cache
 
well, gaming for hours is stable lol. I tested for 30mins for 1.975v bios 1.456windows but i had to turn it off. I'll see how it goes.
 
1.975!?!?!
Must be 1.475 because windows wouldn't report that much lower.
 
wtf? it fails after 30 mins. This is so weird. I was stable for hours before i changed the fans... now I have to add .1v more ? WTF?

I was at 1.46v (before the fan change) now I am at 1.5 and still not stable (for orthos atleast...maybe its an orthos problem?)
 
You must log in or register to reply here.
Back
Top