Dual PIII800 under Win2k Pro intermittently locking up hard

[RolandG]

I've got an Intel OR840 motherboard with two PIII-800 133FSB's, 256MB of PC800 RDRAM, a Quantum Fireball LM30, and a Matrox Millenium G400 video card all running under Windows 2000 Pro. My problem is that the system intermittently will lock up hard. The cap's lock & num lock don't even respond. There is no BSOD from W2kPro & no event is recorded in the system event log. I guess when I say that this happens intermittently I should add that this only happens after the system has been up for about 6 to 7 days straight. The system is always proc'ing two seti@home packets (commandline version client 2.4)& these lockups have always occured while I was activly using the system. These lockups began after I added the 2nd PIII 800 chip several weeks ago. I have re-installed Windows 2000 and the system still locks up after about a week of uptime. Prior to adding the second processor the system had been completely stable running for up to 30 days between re-boots and it NEVER locked up. Anybody have any guesses as to what may be causing these lockups? It seems to me that if there were any major hardware problems I'd have lockups every day. I guess I can live with the problem given the system stays up for about a week but it's annoying the heck out of me. Thanks in advance for any ideas on solving this problem.

 

[brewty]

i'd say either it's your ram or your new cpu isn't compat with your other cpu (they both same stepping, or listed as compat on intel's site?). i say ram because it doesn't crash when you leave it running seti, but when you use it actively, it crashes (pretty much using cpu only when you don't use it, and uses cpu AND memory when you are using it).

 

[RolandG]

Both CPU's are PIII 800 EB's stepping 3 & both are from the same FAB in Malaysia. The new CPU was from an Intel "boxed" set & the other was an OEM cpu. I have been wondering if heat could be an issue on the second cpu because of the crappy intel provided heatsink/fan. According to Intel LanDesk Clientmanager the new cpu runs about 10-12 degrees hotter than the OEM marketed CPU with an AVC heatsink/fan. The room temp here now is about 72 F and according to LanDesk the new cpu is running at 56 degrees C while my original cpu is running at 47 degrees C. This is under a full processor load by the way. According to LanDesk the critical temp. for these chips is 76 degrees C. That seems pretty hot to me. The motherboard is running at 29 degrees C. I have two case fans (one front pulling air in & one back exhausting in addition to the pwr. supply fan exhausting.) and the air in the case never seems to get very warm. Could the second cpu getting up to 59 or 60 degrees C be too hot for it & be causing my lockup problem? When I said in my original post that the system proc's seti all the time & locks up when I'm using the system, I have been using the system for 6-8 hrs a day during the 6-7 days of up time prior the times it has locked up. It doesn't just lock up whenever I start using it (at least it hasn't so far). It NEVER had any problems prior to the 2nd proc so I hadn't really considered the RAM to be a possible source of the problem since I didn't change anything RAM wise when I added the 2nd. Thanks for your help.

 

[Sukhoi]

Hey Roland, are you running SETI@Home for Team AnandTech?

 

[Sukhoi]

Wow, I just noticed and those CPU temps do seem hot, but I doubt they have anything to do with the lockups because the lockups are so far apart. Does your case have good ventilation? My Celeron 433 runs at about 42 C and I thought that was hot. I think Coppermines are supposed to run a lot cooler than .25 micron processors. You might try putting a 120mm blowhole in your case right above the processors...that should help the temps a lot.

 

[RolandG]

I just joined Team AnandTech which should add my 419 work units under my Hotmail account name abusi.

 

[RolandG]

My case does have pretty good ventilation. It has 4 1/2' fans at the front and back with the rear fan located right behind the two CPU's. There's quite a bit of cool air going over the CPU's. This motherboard orients the CPU's so that they are mounted horizontally with CPU #1 above cpu #2 so they seem to be getting equal airflow over them from the case fans. CPU #1 did run about 4 C cooler before I added #2 because the heatsink fan didn't have the other CPU directly below the the fan. The only reason I can figure the Intel boxed CPU is running so much hotter is because the fan in the Intel provided heatsink is about half the size of the fan on my other CPU. If I was to go out and buy a new fan/heatsink for the Intel Boxed chip would it void the 3 year Intel warrantee? It doesn't look like it would be very easy to remove the Intel heatsink.

In trying to figure out what may cause the lockups it seems that each time I've had a lockup the modem was downloading files & I was moving windows on the screen when the lockup would occur. Windows 2000 has put my modem and video card both on IRQ 16. Supposably IRQ sharing should not be a problem under Windows 2000 on an ACPI machine but I can't help but wonder if these lockups might not be as a result of an old fashion IRQ conflict. Although you'd think it would cause a problem in less than a week of uptime and these devices may have been sharing the IRQ prior to my re-install of 2000 which I did after I got my second CPU. Unfortunately with Win 2000 and ACPI I can't manually assign the modem or video card to a different IRQ to see if that would make any difference.

 

[Sukhoi]

Welcome to the team! If you haven't already, you should go over to the Distributed Computing forum and check out the stats. How many WU/day do you get with your computer? My uneducated guess would be about 6 WU/day, but I'm probably wrong..hehe.

You seem to have good cooling in your case, so I can't really think of any way to get your processors cooler without going to water cooling or something wild like that. Yes, replacing the Intel heatsink/fan with something else will void the warranty. Well, I don't know 100%, but I've heard a lot of people saying that and I can't believe they're all wrong.

It's possble you're having IRQ conflicts, but I really don't have any idea as I don't know a lot about Windows 2000. You should try posting the problem over in the Operating Systems forum too.

 

[RolandG]

Thanks for the welcome. I usually complete about 6 or 7 work units a day on this machine. Every couple days my old P5-166 completes a work unit too =) (it takes that sucker about 34 hours per unit). I run two instances of the commandline 2.4 client & each client takes about 6 hours to complete a work unit. I have to run two instances beause the client is not multithreaded and Win 2000 will not split the processor load across both cpu's. The only way to get full utilization out of both cpu's with the command line client is to run the multiple clients. The graphical seti client will use both cpu's but unfortunately takes twice as long to process a packet.

 

[Assimilator1]

Only if you don't set the GUI version to blank screen in 1min or so

Welcome to our team BTW

Just a crazy thought ,but I don't suppose you've tried swapping the cpu's round have you? ,that might help that other cpu cool a bit better.

BTW someone at the Disributed forum might be able to help you ,mention SETI in the thread again & it won't be OT , say hello to the rest of us to