Probably silly question: Do CPUs "age"?

Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

Threaded View

Ok, here's the situation:

Two PCs, both using the same mobo (Soltek SL-75FRD2-L), nForce2 Ultra 400
chipset. Both have unlocked Barton XP2500+ CPUs. Both around the same age...
being from when Bartons were the (consumer) peak of AMDs CPU line. Both CPUs
overclocked very well, dong 2.2GHz easilly, (200MHz FSB) albeit requiring
slightly different vcore settings to be Prime-stable. One CPU managed it at
default (1.65V) whereas the other needed 1.725V. However, using the
Thermaltake TR2 M3 HSF:

Both ran at under 60C even in a warm (30C) room after running Prime for
hours. Good case ventilation.

Recently both machines have been unstable. (One is mine, one I built for the
ex). Mine had me stumped for a couple weeks. It would go through the BIOS,
get to where it should start Windows (XP Pro SP2) and crash back to the
video BIOS and start again. Sometimes 6 or 7 times before booting into
Windows. I replaced the HDD, reinstalled as I thought it could be the
Ghosted image, swapped on-board IDE, tested +/-5V and +/-12V with a
multimeter while the machine was booting. Raised chipset voltage a notch,
raised VRAM a notch, couldn't work it out. Once in Windows it ran Prime
flawlessly and would boot into Memtest86 and run that all day. I even bought
a PCI-IDE card as I suspected the on-board IDE controller. After much
pulling out of hair and rending of clothing I tried increasing vcore from
1.65V to 1.70V and it was absolutely fine. First time I've ever had a
machine that wouldn't always boot Windows but when it does is Prime stable.
Normally they'll boot into Windows but not be Prime stable...

As pleased as I was to have 'fixed' it I was irked. It seemed odd to me. Has
anyone else seen this sort of behaviour? This was about 4 months ago.
Yesterday I rebooted (It'd been running non-stop for a week) and it did the
same thing. Grrr!!! Raising the vcore another notch to 1.725V has it booting
up just fine.

The thing is, I thought that maybe something's going out of spec and the
actual vcore is staying constant, I'm just having to 'raise' it in BIOS to
maintain the same as I always had. However, MBM5 tells me different. As does
the slight increase in temps.

At about the same time the ex was hassling me to look at her machine, random
reboots and unpredicatble behavior. With hers I found that I was absolutely
unable to get it to run at 200MHz FSB regardless of multiplier. I set it to
run at 166MHz and 12 multi and it ran perfectly at 1.675V vcore. However,
recently it's been randomly crashing and I've had to go up to 1.70V.

I've *very* carefully eyeballed the mobo capacitors and all seems fine.
They've both always ran at cool temps, more than adequate case cooling and
they have different brands of PSU so I can't put the blame there. (Both
520W, a Task and a Thermaltake).

So, what do you think? I thought it was probably mobo caps at first but, as
I say, they're cooled well, kept clean and look pristine. Also, the vcore
increasing in Windows/MBM5 leads me to believe that it's not mobo-realted
voltage sagging. (I check MBM5's high/low log regularly to look for
abberations, none found).

Can it be CPUs? It seems unlikely to me but I find myself a bit puzzled over
this. I built these machines hoping to get around 5+ years or more out of
them (neither of us get into demanding gaming, both running nForce 4
ti4200's). However, the current trend of constantly needing higher vcore is

Any input/advice/theories appreciated.


Re: Probably silly question: Do CPUs "age"?

Thermal compound ages and can become very less efficient over time. CPU
heatsinks get dust clogging causing lower performance. Check the
motherboard's capacitors as they seem to not last very long. Most
instability I have seen was caused my RAM and RAM timings.

Re: Probably silly question: Do CPUs "age"?

Kent_Diego wrote:
Quoted text here. Click to load it

The CPU temps as reported by the on-die termistor via Motherboard Monitor 5
have stayed consistant (in relation to ambient/case temps, (except for going
up a degree or two after raising vcore)) so I don't think it's a TIM
problem. My heatsinks (and all case fans) are cleaned religiously every few
months and the RAM on both machines is set to "By SPD" and has been
repeatedly tested with a Memtest86+ bootable CD without it reporting any


Re: Probably silly question: Do CPUs "age"?

Quoted text here. Click to load it

Yes, this does seem like a familiar story. I've read of some Athlon64's
doing this when overclocked. The owners note that the max overclock
drops and drops, until eventually the processor won't even run at
stock speed. (At which point they sold the processors on Ebay.)

The only wearout mechanism I've heard of is electromigration. Basically
this is a problem with the conducting wires on the silicon die.

   "In modern consumer electronic devices, ICs rarely fail due to
    electromigration effects. This is because proper semiconductor
    design practices incorporate the effects of electromigration
    into the IC's layout."

I doubt this is a problem with your motherboard.


Re: Probably silly question: Do CPUs "age"?

Paul wrote:

<snip tale of woe>

Quoted text here. Click to load it

Thanks Paul. I've not heard of this phenomenen in connection with CPUs
before and have in fact run (for instance) two Celeron Coppermine 600's at
900MHz for years without having to raise vcore after finding the initial
'optimal' level with no problems. (They're still in my drawer in fact).
Perhaps it's just a design flaw in (certain) AMD CPUs.

I'd be interested to hear more of this phenomenen if that's the case. I have
an XP1800+ T'bredB (JUIHB) in an identical mobo that is running well O/Ced
(2.1GHz on a 200MHz FSB instead of specced 1.53GHz on 133MHz FSB. It seems
to do twice the work, mostly due to the FSB increase methinks) that hasn't
exhibited this sort of behavior at all yet I've had it over a year longer.


Re: Probably silly question: Do CPUs "age"?

~misfit~ wrote:
Quoted text here. Click to load it

<snip rest>

Typo, SL-75FRN2-L. Sorry.

Re: Probably silly question: Do CPUs "age"?

~misfit~ wrote:
Quoted text here. Click to load it

I've the same setup doing the very same thing.  It's currently stable at

Soltek sent me capacitors to replace the ones that have bulged, but
sweating copper pipes is about the extent of my soldering expertise. :\

The world's least questioned excuse for missing work....
"the voices in my head told me to stay home and clean guns today".

Re: Probably silly question: Do CPUs "age"?

pheasant wrote:
Quoted text here. Click to load it

Hey pheasant,

What country do you live in? You had an SL-75FRN2-L that had capacitor
problems and was doing the same as mine are? Can you give me info as to how
you went about getting in touch with Soltek etc please? I can be reached at
misfitnz at gmail dot com if you'd rather take it to email.


Site Timeline