mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware

Reply
 
Thread Tools
Old 2005-01-21, 10:00   #1
beorntheold
 
beorntheold's Avatar
 
Jan 2005
Germany

7 Posts
Question Prime-Unstable...

Hey guys *waves*,

A couple of days ago a guy I know formed a folding team and I wanted to join. The f@h console froze my "otherwise stable" machine a couple of times and, after doing some reading, I ran Prime95 to check for instabilities.
It froze once, most of the times it just reports errors after a couple of minutes at default clocks (133) with every working memory timing . So I ran memtest. To my surprise it ran okay for several hours (5-6 hours) at the Probe setting...

Some additional info about my system *points down at the end of the post*:
- running XP Pro SP2 with all updates

- the mobo (bought second hand from ebay a couple of weeks ago) looked brand new - according to the previous owner he used it only a couple of times; the only things I can change are FSB and memory timings; the monitoring doesn't work and I know how to set up MBM - the only readings that make sense are CPU temp (~47-53C) and vcore (stable 1.60V) - all other voltages vary wildly; the vcore reported by the BIOS and CPU-Z is 1.632V - stable; the northbridge used to get pretty hot, so I put a small 40x40 fan on it - it should be ok now.

- the CPU (bought new a couple of weeks ago as well) is a T-Bred "B" 2600+ (16x133) and has been OCed only a couple of MHz (135), mainly to avoid the feeling that the machine is underclocked, which, as most of you probably know, is a thing the guys from ECS have the habit of doing...

- the PSU (bought new at the same time) is a 550W LC-Power and the voltages reported by MBM on my old mobo were very close to optimal and absolutely stable; the stable vcore on this one makes me think the problem is elsewhere

- the RAM (I've had it for a while) - 2 x 256 Hynix, SPD @ 133 2.5-3-3-6, tested with memtest @ 2-3-3-6

- the video card - a 9800 Pro, bought through ebay, supposedly manufactured by Hercules (has logos and markings on the typically colored PCB that support the theory...) - a week or two after I bought it I saw some memory-caused artifacts and looked at the card more closely - the thermal paste under the heatsink looked like it was applied by a half-trained monkey with it's left leg; the heatsink looked just like ATI's stock solution, but wasn't from aluminum(!?); the 350Mhz Samsung memory used to get so hot that I got scared and changed the heatsink and I run the card slightly underclocked now. I haven't noticed any artefacts since, but I don't really trust the card.

- the sound card (bought second hand through ebay - didn't want to use the onboard codec) - SB Live! ...

Well, that's pretty much it. Sorry for the novel and thanks in advance for your ideas! :)

P.S.
- flashing the latest BIOS version didn't help
- here's a part of the error messages I've been getting:

...
ERROR: ILLEGAL SUMOUT
Maximum number of warnings exceeded.
[Thu Jan 20 17:16:37 2005]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Fri Jan 21 02:42:09 2005]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Fri Jan 21 05:53:32 2005]
FATAL ERROR: Rounding was 0.499710083, expected less than 0.4
Hardware failure detected, consult stress.txt file.
ERROR: ILLEGAL SUMOUT...


--------------------------
Athlon 2600+ T-Bred "B", ECS N2U400-A r.1.0, 512 Mb PC266 Hynix, Hercules 9800 Pro, 550W LC-Power

Last fiddled with by beorntheold on 2005-01-21 at 10:07
beorntheold is offline   Reply With Quote
Old 2005-01-21, 12:13   #2
Mystwalker
 
Mystwalker's Avatar
 
Jul 2004
Potsdam, Germany

3·277 Posts
Default

I'd try each memory module individually - if one configuration runs and the other doesn't, we have a suspect.
Apart from that, take out the sound card. Sometimes, faulty drivers make the system a bit "fuzzy".
Mystwalker is offline   Reply With Quote
Old 2005-01-22, 06:17   #3
beorntheold
 
beorntheold's Avatar
 
Jan 2005
Germany

7 Posts
Default

Thanks for the quick response!

Quote:
Originally Posted by Mystwalker
I'd try each memory module individually - if one configuration runs and the other doesn't, we have a suspect.
Apart from that, take out the sound card. Sometimes, faulty drivers make the system a bit "fuzzy".
Tried each module individually - rounding errors and illegal sumouts in first Blend test after only a minute or two...

Took the SoundBlaster out -

Checked for damaged capacitors - as far as I can see there are none.

Small FFTs test ran okay for about 10 minutes then it caused the system to restart. I'm starting to think that MBM may be right and the BIOS reading is wrong, that the vcore is actually 1.60V. According to my research the board doesn't even have the circuitry to change the vcore per BIOS... and I really don't want to do a pin mod...
beorntheold is offline   Reply With Quote
Old 2005-01-22, 10:24   #4
beorntheold
 
beorntheold's Avatar
 
Jan 2005
Germany

1112 Posts
Default

Quote:
Originally Posted by beorntheold
... the monitoring doesn't work and I know how to set up MBM ....

I had overlooked something when I configured MBM...
Now I have voltage readings. CPU - 1.63V. The Core 1 reading (most likely a 2.5 V line) varies between 1.7 - 3.0 V, pretty unstable. All other readings (+3.3, +5.00, +12.00, -12.00, -5.00) are relatively stable an within limits.
beorntheold is offline   Reply With Quote
Old 2005-01-22, 12:01   #5
beorntheold
 
beorntheold's Avatar
 
Jan 2005
Germany

7 Posts
Default

Another update:

I think I might have localized the problem - under heavy load (Atitool, prime's Small FFT's test) the CPU voltage drops under 1.62V - CPU-Z reported 1.616V...

Last fiddled with by beorntheold on 2005-01-22 at 12:04
beorntheold is offline   Reply With Quote
Old 2005-01-22, 18:17   #6
beorntheold
 
beorntheold's Avatar
 
Jan 2005
Germany

7 Posts
Default

Did a pin mod to 1.70V to compensate. Voltage stays at 1.65-1.66V under load! Prime95 is running TWENTY MINUTES without errors!

Last fiddled with by beorntheold on 2005-01-22 at 18:20
beorntheold is offline   Reply With Quote
Old 2005-01-23, 10:23   #7
beorntheold
 
beorntheold's Avatar
 
Jan 2005
Germany

7 Posts
Default

Final update:

folding at home console has been running for 14+ hrs now - case closed...
beorntheold is offline   Reply With Quote
Old 2005-01-23, 14:58   #8
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

175708 Posts
Default

Tell us about your avatar... It looks interesting...
Xyzzy is offline   Reply With Quote
Old 2005-01-23, 19:04   #9
beorntheold
 
beorntheold's Avatar
 
Jan 2005
Germany

7 Posts
Default

Quote:
Originally Posted by Xyzzy
Tell us about your avatar... It looks interesting...
Visit http://www.tokyoplastic.com/musicboxsplash.html or, even better, start at http://www.tokyoplastic.com. Artistically, it's one of the most stunning sites the Internet has to offer. Imho, that is...

Oh, and - be sure to check out the drum machine!

Last fiddled with by beorntheold on 2005-01-23 at 19:09
beorntheold is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
(M48) NEW MERSENNE PRIME! LARGEST PRIME NUMBER DISCOVERED! dabaichi News 571 2020-10-26 11:02
disk died, prime work lost forever? where to put prime? on SSD or HDD? emily PrimeNet 3 2013-03-01 05:49
How do I determine the xth-highest prime on prime pages? jasong Data 7 2005-09-13 20:41
Unstable? Unregistered Hardware 10 2005-01-15 18:17
The 40th known Mersenne prime, 220996011-1 is not PRIME! illman-q Miscellaneous Math 33 2004-09-19 05:02

All times are UTC. The time now is 00:09.

Wed Apr 21 00:09:55 UTC 2021 up 12 days, 18:50, 0 users, load averages: 3.59, 3.40, 3.52

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.