Athlon XP-M 2500+ failing Prime95

[Pollock]

Well, here's the story. A friend wanted me to upgrade his socket A system. It doesn't need to be able to play games or anything.

Before:
1800+ Palomino + generic HSF
MSI K7N2 (nF2 supporting 400FSB)
256MB PC2100
Dying Maxtor 60GB PATA
Some generic PSU (9A on the 12V+ )
GF4MX
SB Live! 5.1
CD-RW

Upgraded to:
2500+ Barton Mobile + generic HSF
K7N2
2x512MB PC3200
Seagate 160GB PATA
Skyhawk 385W (not the greatest, but 16A on the 12V+)
GF4MX
SB Live! 5.1
DVD-RW and CD-RW
PCI modem

This thing failed Prime95 in under a minute at 200*8=1600MHz and 1.6V. At the moment it's running Prime95 again, and I haven't seen temps much above 50C over the course of several hours of trying to figure this out. The monitor turns on and off at times, and occasionally there are artifacts on the Windows desktop (only when running Prime95 though). Something is wrong here, but I can't seem to figure out what. I've tried another PSU that I know for certain should have no problem with that system, and still had problems. It was Prime stable for 12 hours with the 1800+ before I swapped it out. I've updated the BIOS and reset the CMOS a couple times. I've done some other stuff, but nothing successful. Any other further suggestions to try?

 

[Slikkster]

Shouldn't be a problem OC'ing this with your system, according to http://www.neoseeker.com/Articles/Hardware/Features/XP-M. You're easily withink OC range on this cpu.

I would tend to look at a couple of things. First, pull one of the sticks of your RAM and see what happens. No change? Swap and pull the other one. Don't know what memory timings you're trying to use, but you might want to lower them.

Also, I'd double check to see if I did a good job with thermal paste and contact with the heatsink on the cpu. And, I'd use CPU-Z to make sure I really had a mobile Barton, vs. some altered one passed off as such.

 

[Pollock]

I already checked with CPU-Z to confirm a few things, and yes, it is a mobile Barton. I'm running the ram at SPD, although it won't hurt to run memtest during the day. I was going to check the HSF contact, but I couldn't get it back off and gave up. I hate removing socket A HSFs. I'll try that again after running memtest.

 

[Fardringle]

Along with the steps that Slikkster suggested, I would try running the CPU at its stock settings to make sure it is stable at that point, and then go ahead and drop the multiplier and put the FSB back to 200 after you confirm that the CPU and motherboard are working properly together.

 

[Slikkster]

I wouldn't just rely on memtest. It's pretty easy to just swap out the sticks one at a time and see how the system responds.

By the way, if there's no gaming involved, why is it so necessary to OC this thing? Microsoft Word is only going to open so fast, if you catch my drift.

 

[Pollock]

Originally posted by: Slikkster
I wouldn't just rely on memtest. It's pretty easy to just swap out the sticks one at a time and see how the system responds.

By the way, if there's no gaming involved, why is it so necessary to OC this thing? Microsoft Word is only going to open so fast, if you catch my drift.

Well, he's giving it to his niece, but still wants it to be plenty fast and not need to be upgraded again. I mean, it's not like doing 2GHz on a mobile Barton is really overclocking it or anything. That should be guaranteed. That said, it should be stable at 1600MHz with 1.6v no problem, with the FSB at 200MHz. Matter of fact, 1600MHz is probably underclocking it a little.

It passed memtest with both sticks for 12 hours, but I went ahead and tried it with only one stick at a time. I first removed the stick from slot 3 (if that makes any difference) and it just passed its first iteration of FFTs from 2048 to 4096. Hopefully it is stable now (for whatever reason).

By the way, when I thought it was successfully priming the second time earlier...it was actually frozen. :frown:

 

[Pollock]

Spoke too soon. It just rebooted. :frown: Time to swap the other stick...