• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Uh-oh... CPU degrading?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Dooms101

Member
Joined
Dec 28, 2009
Location
under a heatsink
Well... last night around 1:00am I was up trying to get a Mac OS X virtual machine running so I could run Xcode when suddenly I BSOD'd while extracting a large iso .rar archive. My computer has been running stable for a month or two with these settings:

Bus: 206.3mhz
CPU: x15.5 (3.198ghz)
vCore: 1.34 - 1.35 (my board likes to sporadically throttle)
RAM: 3:8 1100mhz
Timings: 5-6-6-18-24-2t (SPD timings for 1066mhz)
vDIMM: 2.09v (1066mhz rated at 2.1v)
CPU temp: ~51C max load (49C average load)
NB temp: ~26C load

These settings have been working flawless for everything and passed 4 hours large Prime95. I've gotten my CPU benchable at 3.7ghz and my ram stable at 1120mhz with 6-6-6-18-32-2T. So, these settings seemed very safe. However, after my BSOD I couldn't boot. So I reset the CMOS like always and after a few reboots into BIOS I had my O/C back up. It got to the Window's loading screen and crashed followed by a reset. I tried setting the vCore to 3.6v and still crashed. I reset the CMOS again and ran stock settings perfectly fine.

This morning, I rebooted and tried my OC again with the same result, it just wont boot into windows now. There's no jump in temperature at all, and nothing in my whole system gets hot. I now have it running at:

Bus: 200mhz
CPU: x15 (3.000mhz)
vCore: 1.31v
RAM: 1:2 800mhz
Timings: 4-5-5-15-24-2T
vDIMM: 1.9v

What do you guys think is happening? This system is ran 24/7, and is cleaned every 2 weeks. The parts were run stock from when I first bought it 6 months ago up until 2 months ago. And I've had a stable OC for about a month and a half. Before I cut a top 120mm fan mount (see mod wip thread in my sig for details) the PSU would get warm after an hour of use, but now it's always nice and cool. The power phases on my board don't have a heatsink on them and get pretty hot when really OC'd like when using 3.7v or more. I think they're the problem, but they don't seem that hot, I'd estimate around 40C when using my 3.4v OC. The NB used to get hotter then I like, but after I got my new case mod done, it's been running around 26C load with 20C ambient. I have a feeling it's something to do with power, so either my old Ultra PSU is going, or my board is. Do you think it could be something else?
 

DragoXT

Member
Joined
Jun 14, 2005
try running a chkdsk to make sure there wasnt something with the disk that caused the BSOD. Then put your cpu back at stock speeds but have your RAM back to your 24/7 OC that failed and run memtest and see if it fails. Maybe just maybe you didnt have enough voltage in the RAM at those settings and when stressed hard it could fail.
 
OP
Dooms101

Dooms101

Member
Joined
Dec 28, 2009
Location
under a heatsink
Well I think what did it was because I accidentally sent the iso .rar to my desktop instead of a 'normal' directory. Windows 7 has some bugs with normal file operations with the desktop. So I know that's what caused the BSOD. It's just the fact that it wouldn't even boot after setting the 24/7 OC back up when I reset the CMOS.

What you said seems like a good idea, I am pretty sure it's my PSU that's failing since the temps haven't changed.
 

AlabamaCajun

Member
Joined
Apr 22, 2005
Location
A Labyrinth
You can get Prime 95 running for days and kill the same proc with folding at home or other apps. As good as these torture testers are, they can still miss opportunities that code paths can take that sink an otherwise solid rig. If anything is really going to degrade, the MOBO is prime target in the VRegs. This is more common with non digital VRegs. Heat is probably the main reason for degs.
 

QuietIce

Disabled
Joined
May 7, 2006
Location
Anywhere but there
You can get Prime 95 running for days and kill the same proc with folding at home or other apps. As good as these torture testers are, they can still miss opportunities that code paths can take that sink an otherwise solid rig. If anything is really going to degrade, the MOBO is prime target in the VRegs. This is more common with non digital VRegs. Heat is probably the main reason for degs.
I've seen people scoff because one of my rigs can't run P95 overnight (it'll run OCCT 24 hours) but that same rig has been running SETI for a solid year now without problems. Sorry, I'm gonna' call that one stable and move on - I don't care what P95 is doing to it ... ;)
 
OP
Dooms101

Dooms101

Member
Joined
Dec 28, 2009
Location
under a heatsink
lol... well I have it running 24/7 with:

Bus: 200mhz
Multi: x15.5 (3.100ghz)
vCore: 1.310v
Ram: 800mhz
Timings: 4-5-5-15-22-2T
vDIMM: 1.86v

I am not really sure whats going out on my system, but I am guessing I can keep it alive for a little bit longer with these settings and still keep a good OC, I am planing on a new build soon enough anyways.
 

Niku-Sama

Member
Joined
Jan 13, 2005
I've seen people scoff because one of my rigs can't run P95 overnight (it'll run OCCT 24 hours) but that same rig has been running SETI for a solid year now without problems. Sorry, I'm gonna' call that one stable and move on - I don't care what P95 is doing to it ... ;)

yea i agree with this, i have had the same thing, fails prime but is able to run more intense applications just fine for long periods of time.

i like OCCT better any way and i like how it has a bunch of other tests in it now
 

QuietIce

Disabled
Joined
May 7, 2006
Location
Anywhere but there
Whatever is going out on your rig I seriously doubt it's the CPU. There are some failures straight out of the factory but once they're running they just don't fail without some type of serious physical damage. MOSFETs by their very nature wear out over time as do other system components but most CPU-type chips just don't go bad from time alone ...