Opty 170 rig bluescreens StressPrime 2004

[Aquila76]

How reliable is the Orthos StressPrime 2004 (Beta dual-core version)? Every time I run it, it runs great for a while, then I get a BSoD. I've had this system pretty stable until I tried to install the 1805 BIOS, but I believe successfully recovered from it.

Current setup:
A8N-SLI Deluxe, BIOS 1017.004
OCZ EL Platinum Rev. 2 1GB (2x512MB) @stock
Opty 170 retail (Model: OSA170DAA6CD, Stepping: CCBBE 0615EPMW) w/ Zalman CNPS 7000-Cu @ stock

I just upgraded from my A64 3500+ to this Opty 170. All I'm trying to do is setup/burn-in the Arctic Silver 5, not OC'ing anything yet. I did install the following from AMD's site in this order:

1.) AMD Opteron? Processor with AMD PowerNow!? Technology Driver for Windows XP and Windows Server 2003 Version (x86 and x64 exe) 1.3.2.16
2.) AMD Dual-Core Optimizer
3.) AMD Power Monitor Version 1.0.1

Any ideas?

 

[corsa]

So i take it u want the temps to run higher than usual for awhile...unplug the 7000-Cu and use as per normal, no need to stress with prime. On a side note i've never been a fan of Prime, Gaming sorted my pc out alot earlier than prime ever did.

 

[aka1nas]

is there a driver dll associated with that BSOD? Might be an Nforce driver issue. I definately had my share with the A8N series.

 

[Aquila76]

Originally posted by: aka1nas
is there a driver dll associated with that BSOD? Might be an Nforce driver issue. I definately had my share with the A8N series.

The BSoD I saw didn't specify any specific driver. Just the generic "If you've added any new hardware or drivers..." message. Interestingly, I tried re-running SP2004 again in hopes to get the complete BSoD language, and got this error message (apparently on Core1 (top window)):

Self-test 16K passed!
Test 1, 270000 Lucas-Lehmer iterations of M420217 using 20K FFT length.
Test 2, 270000 Lucas-Lehmer iterations of M409601 using 20K FFT length.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
Torture Test ran 1 hours, 31 minutes 47 seconds - 1 errors, 0 warnings.
Execution halted.

The splog file shows that Core1 is the last entry when the system must have BSoD'd. Crap, I hope I didn't somehow zap my new chip when installing it. I'm going to re-check the BIOS settings to make sure everything is set right.

 

[myocardia]

Originally posted by: Aquila76
Crap, I hope I didn't somehow zap my new chip when installing it. I'm going to re-check the BIOS settings to make sure everything is set right.

Just raise your vcore to 1.35v, where it's supposed to be, and your problems will go away. Well, assuming your motherboard doesn't undervolt by much. You may have to go as high as 1.40v, to counteract vdroop.

 

[Aquila76]

Originally posted by: myocardia

Originally posted by: Aquila76
Crap, I hope I didn't somehow zap my new chip when installing it. I'm going to re-check the BIOS settings to make sure everything is set right.

Just raise your vcore to 1.35v, where it's supposed to be, and your problems will go away. Well, assuming your motherboard doesn't undervolt by much. You may have to go as high as 1.40v, to counteract vdroop.

Thanks for that; it was autodetected to 1.30V. I noticed that for some reason, the BIOS was pulling the RAM timing from SPD wrong, too (should be 2-2-2-5, pulled as 2.5-3-3-11). I just manually set everything (Voltages, speeds, multis, etc.) to where it should be. I'll check back on it tomorrow morning.

 

[myocardia]

No problem. If your BIOS has a setting for 1.375v vcore, I'd go with that, since nearly every motherboard ever made has at least .025v vdroop. Also, aren't your RAM timings are supposed to be 2-3-2-5, not 2-2-2-5?

 

[noping]

Originally posted by: Aquila76
How reliable is the Orthos StressPrime 2004 (Beta dual-core version)?

Fairly. It's just a modified version of Prime95. I prefer it over trying to mess around with Prime95's settings to get it to run on both cores.

Originally posted by: myocardia
Also, aren't your RAM timings are supposed to be 2-3-2-5, not 2-2-2-5?

Those Rev. 2 TCCD/TCC5 sticks are rated for 2-2-2-5 out of the box/stock, I recall (I have a pair).

I had to raise vcore up to around 1.4v on my X2 3800+ (stock speeds) just to pass a day's worth of SP2004...

 

[myocardia]

Originally posted by: noping

Originally posted by: myocardia
Also, aren't your RAM timings are supposed to be 2-3-2-5, not 2-2-2-5?

Those Rev. 2 TCCD/TCC5 sticks are rated for 2-2-2-5 out of the box/stock, I recall (I have a pair).

I had to raise vcore up to around 1.4v on my X2 3800+ (stock speeds) just to pass a day's worth of SP2004...

Ahh, I missed the Rev. 2. Just skimmed, and saw OCZ Platinum, which are 2-3-2-5 Infineon IC's, and not TCCD's. And yeah, it all depends on the board, it seems. Even my DFI board required me to set it to 1.375v, although it was passing Orthos with 1.35v in the BIOS.

 

[Aquila76]

Thanks for the Vcore tip, myocardia; that appears to have been it. It ran overnight (~9 hours now) error free. Temps are pretty stable, too (~43C according to my fan controller's temp probe on the side of the CPU).

That was almost starting to freak me out. I think that's the first time my system's bluescreened since I tried running the nVidia firewall a few years ago (that got uninstalled quick).