(Another) CPU degradation Thread

[ChrissyT88]

Sorry for another degradation thread, but i think i may be experiencing something similar to Vertigo1. I've had a Q6600 since last July which is showing signs of being unstable. When i first got the processor, i clocked it to 3.3 Ghz with a vcore of about 1.35-1.37 i think (VID 1.325). I had quite a few stability issues a few months later and concluded that it probably wasnt all that stable, so using LinX (Linpack stress tester) i got it stable using 1.4v vcore. It has sat at this setting since late november or early december.

Recently has shown signs that i cannot hold the clock - i have been getting a few BSODs, so last night i ran the OCCT large memory stress test for an hour, and it passed fine, as did a 3hr run of memtest, but LinX would not run stably. I have bumped the vcore up a few notches (from reading around, vcore is the likely suspect for Linpack failures) and will see if this improves things. I'll also try and run a long prime lend test. However, i do think its strange that it can no longer hold the clock. It has been running Folding@home for at least 6 hours a day since December, and when its on its at 100% load and has shown no signs of instability (temps also never get above about 59C). What are people's opinions - does it look like the chip is dying?

I've noted my BIOS settings in case there is something obvious i should change. Im not that experienced at overclocking, but from reading around my settings seemed fine:

ASUS P5Q Deluxe Motherboard
Q6600: 367x9
vCore: 1.4000
PLL: 1.52
FSB Termination: 1.24
NB: 1.20
Memory: 1104 Mhz (Corsair Dominator pc8500 RAM), 5-5-5-15 timings, PL of 8, everything other timing on AUTO
Memory Voltage: 2.12

Everything else (Southbridge, GTL etc has been left on AUTO). As i say, it seemed fine until a few days ago, but now is a little flakey...

Cheers, any help is appreciated. If it would be easier to have this in the other thread, i'll close the thread and move it, but i dont want to clog up Vertigo's thread!

 

[gurusan]

Try running memtest to see if the memory is shot.

I've not ever heard of a Q6600 dying from anything less than insane volts with LN2....on a normal clock I guarantee your chip is not degraded.

 

[Caracus2k]

Do some research into 'Electromigration', which occurs much more rapidly at higher voltages. If this is your problem then you will be trapped in a vicious circle of needing higher voltages to keep a stable clock which will speed up the degradation till the chip is dead for good

 

[ChrissyT88]

Thanks gurusan - i ran memtest a few days ago for three passes and did not find any memory errors. I realise that three passes is not ideal, and I will try again this evening and try downclocking the memory to see of that helps at all.

Caracus - i am aware of the problems of electromigration. However, i do find it surprising that it would happen so quickly with voltages that are relatively modest compared with those other users on here are using. For now i have dropped the clock (have work to do!) and will continue testing later on.

Thanks for the replies.

 

[Ben M]

This is the first thing I've heard of a 65nm CPU suffering from degradation at less than 1.4V as well. Very strange.

 

[Happy]

That's very low vcore to be experiencing CPU degradation with

*Looks at his own vcore... *

 

[setter]

Yep strange for such a low voltage, i ran mine for ages on 1.5vcore at 3.8ghz, without issue, only recently dropped down to 3.6ghz at 1.4 vcore for daily use, as 3.8ghz was only showing gains in benchmarking.

 

[Geckovich]

Are you sure the instability isn't due to the fact that we've gone from snowy and cold to quite pleasant and sunny in the space of about 4 days?

 

[jmc007]

^^ QQ

 

[ChrissyT88]

I dont think so - as i mentioned, CPU temperatures dont get abive 60C despite many hours of operation, and are more typically around 55-56 (unfortunately, the quad is remotely flat, so there is quite a difference between the cores, but the other two are typically at 49-51). It will run the OCCT stress test/OCCT Linpack where the cores get to well over 70C without the PC crashing, until the cores get to 80 when OCCT shuts itself down. I guess the memory could be overheating, but airflow through the case is good. Im open to suggestions if the heat could be affecting the system in another way...?

I thought it could be the northbridge/chipset overheating, but Everest reports a motherboard temperature of 31C at the moment, which isnt too hot.

 

[Happy]

What cooler does it have atm?

 

[ChrissyT88]

It has a TRUE with a 120mm Noctua fan. The case is a Lian-Li a17 with a two front 120mm Xilence fans, a 120mm Xilence fan as the exhaust and a 140mm yate loon fan in the roof of the case as an intake.

 

[ChrissyT88]

Mmm... had a bit of time to have a play today. For the past few days i downclocked it to 3Ghz and dropped the memory to just over 1000MHz, and everything has been right as rain. I fired up the PC today with the processor at 3.3Ghz, 1.4 volts as before, but lowered the meory clock and it ran through 10 loops on LinX, reaching a peak of 78C with no crashes, no computational errors. I rebooted and increased the memory clock to 1104 (as it was before) and ran the linpack test again, 10 loops, peak of 79C and no crashes/computational errors. At the weekend, it wouldnt complete more than 2 lopps of linpack at these settings...

The only thing that has changed is the location. I'm back from uni for the holidays and my room here is a bit bigger and cooler, and of course the weather is a little less hot. I am therefore forced to conclude that this has all been an issue with temps...? Im not sure what other peoples opinions of this are - during normal operation, the processor never reches temps higher than 58/59C and the case is well ventilated. I'll run a few loops of memtest, although it has previously passed three with no problems... Mmm...