Wierd Prime95 Problem

[Kolath]

Okay, I've been having a weird problem with Stress Prime 2004. I just got this rig last week and have been trying to overclock:

AMD64 Venice 3200, LBBLE 0516DPHW
DFI UT NF4 Ultra-D, 4.14-2 BIOS
2x512 OCZ Value VX
PowerColor Radeon X800XL 256MB
OCZ PowerStream 520W
XP-90 + Panaflo 90mm fan

I've been trying the method of testing the maximum of each component (HTT, CPU, Memory) so that I can see how far I can push the CPU and board. So far I've been able to get the following:

HTT Max: 285 with CPU@1.6V, mem 1/2 divider, 7x multiplier, 2.5 HT Multi
CPU Max: 2550MHz with CPU@1.6V, mem 1/2 divider, 10x multiplier, 2.5 HT multi
Max Mem: Haven't gotten that far yet, still working on CPU

So I've been having a wierd problem when stability testing. I ran Stress Prime 2004 for 10 hours with no errors while running 8M Super Pi calculations at the same time. CPU never got above 40C. But then when I tried to end the Stress Prime testing, I got a PFN_LIST_CORRUPT blue screen. :confused: Since then this has happened several times, it will not give any errors while running but will crash the computer when I stop running the tests and give me either the PFN BSOD or the IRQL_IS_LESS_THAN BSOD.

Any idea why that might be? It runs (and stops) fine at 2500 MHz or less. Help would be greatly appreciated!

EDIT: Also this is my first post on these forums (have been reading for the last several weeks) and the first system I've ever built completely by myself! :D

[shuRe]

:welcome: Welcome to the Forums! :welcome:

thats a wierd problem there that i havnt heard of (i think it might be somin to do with the memory but dont put me to that). Maybe try some of the following:

-reset the cmos
-lock pci and agp slots if you already havnt
-try an older version of prime95
-reinstall windows (i know its a pain)

if all else fails just dont use prime95, its for extreme stability which isnt allways needed. Try running Super PI 32m, and other stability tests such as looping 3dmark 05 (although im not so sure how harsh this is on the cpu compaired to the gfx card). You could also try occt, which i seen people talking about.

[Kolath]

Thanks for the welcome!

PCI lock is on by default and have tried two other BIOS revisions. I have also reinstalled windows several times (not much of a pain when you have a custom install CD courtesy of nLite :p: ). I will, however, try a different version of Prime when I get home.

Has anyone else ever heard of this problem?

Also, I find that Motherboard Monitor5 and CPU-Z both report core voltages that are about 0.05V lower than the BIOS says they should be. Is this normal? And if so, which monitor is correct?

Also, any general suggestions for overclocking this? 2550 MHz seems kind of disappointing for the new Venice cores. Am I doing something wrong?

[Eldonko]

Sounds like dividers are not stable, its a common problem for this MB and UTT. Thats the BSOD I get when I try and run a divider with UTT, no dividers work at all for me. Find your max RAM speed, then try and run the RAM at the same speed using a divider. This will give you an indication of whether or not its divider related. I'm pretty sure the BSOD is mem related though.

Use the BIOS for vcore, windows programs report it low.

[ChkDsk]

I had the same problem with my Venice... Upgrading to 5-10 bios did the trick. :banana:
It only appeared when running at 1T, 2T worked fine with older bios'...

[brandinb]

dont run super pi while priming that doesnnt stress it more it takes the cpu from prime which stresses it less just run prime alone next time

[TMM]

if all else fails just dont use prime95, its for extreme stability which isnt allways needed. Try running Super PI 32m, and other stability tests such as looping 3dmark 05 (although im not so sure how harsh this is on the cpu compaired to the gfx card). You could also try occt, which i seen people talking about.

on a similar (yet somewhat unrelated, lol) note, i've found Need For Speed Underground 2 to stress my system more then Prime95
:fact:

[Kolath]

Okay guys, you were definitely right about it being a RAM related problem. Last night I increased the RAM volatage to 3.2V (max w/o using the special 4V jumper) and lowered my timings to 3-3-3-7 2T. I was able to get the RAM up to 240 at 1:1 divider. So I tried 2600MHz with a divider and it had no problems in prime. I left it priming over night at 10x270 with the 4/5 divider (RAM running at 225MHz) and it it had errors after 7 hours, but no BSOD!!! I lowered the divider one more notch before work today so I'll see how it handles when I get home tonight.

I will also try the 5.10-2 and 5.10-3 BIOS to see if that can help my memory performance. Thanks a lot for the help! I will report back when I have more results.

[Hoot]

dont run super pi while priming that doesnnt stress it more it takes the cpu from prime which stresses it less just run prime alone next time

Right-on!

Hoot

[CandymanCan]

Thats a common problem with athlon64's happens to me also, why i dont use prime95 much anymore.

[Kolath]

Thats a common problem with athlon64's happens to me also, why i dont use prime95 much anymore.

Which is a common problem? Getting BSODs with prime but seemingly stable otherwise? Do you think its a matter of the A64 architecture?

Oh and from now on I will do Prime or Pi, but not at the same time. Is there a way to run Pi in a loop? I did 32M last night no problems but perhaps it needs to run several times in a row to really stress things?