Do CPUs "age"?

K

_Korruption_

Supreme [H]ardness
Joined
Aug 22, 2002
Messages
4,572
I've had my Locked 2500+ running strong at 2.2GHz (3200+) for well over a year now (specs in sig). It has always needed 1.70V to be able to pass the standard 24hr Prime95 torture test. Recently, my system has begun to reboot itself under heavy load, like in CS:Source.

I thought it was the power supply (Antec True430). It was too noisy anyway, so I swapped it out for an OCZ Powerstream 420W. Great PSU, extremely quiet, looks great, and the rails are spot on thanks to the adjustability. Now, I expected things to get back to normal... but they didn't. I then ran the standard Memtest86+ test overnight to see if it was my TCCD acting up. I doubted it, because most TCCD is pretty solid. Sure enough, my RAM was clean.

So I've ruled out my RAM and my PSU. I'm starting to think that it could be my CPU because now, I need 1.775V just to keep it stable. I originally tried 1.75V, but that wasn't enough. The system still rebooted itself every once in a while. Oh, and it still fails Prime95, but after 19 hours, even with 1.775V.

Is it time for a new CPU? Time for a 2500+ Mobile? I'm not exactly a hardcore overclocker, but being able to run 2.2GHz with "only" 1.65V would be nice. I'd like to get back to the [H] Folding team, but I can't when my computer is unstable! :(



Cliffs:
1) Computer starts rebooting self under load (lightly overclocked, was always stable previously). Now fails Prime95.
2) Swapped in new PSU. Still get the random rebooting.
3) Memtest86+ comes out clean.
4) Starting to wonder if my CPU is flaking out. Needs excessive voltage just to run 2.2GHz, even then, still not stable.
 
They can 'age' in the sense that overclocking (more to the point, prolonged over-volting) can cause problems with electromigration and other effects that can shorten the useful life of a processor. It could also be a motherboard issue.
 
Could be either the CPU or the motherboard. You're not really beating on the motherboard, though, that's a low voltage to be pushing. Throw in a Mobile 2600+ and go nuts.
 
If it's a motherboard issue, how would I go about testing for that? Prime95 tests the CPU subsystem, Memtest86 tests the memory subsystem, but what would test the motherboard?

Oh, and I do know that lower wattage Mobile CPUs are better, but why a 2600+M instead of a 2500+M?
 
my advice would be to make sure there is no dust or crap in your case, make sure its all nice and clean, all fans clean and stuff.

then reseat your processor.

clean the old thermal grease off your processor and apply some new AS5 onto it and reassemble.

thermal grease goes manky after its been on a chip for a while, so best to reapply after a year or so.

all this should lower the temps a bit and get you back to 1.7v.
 
My case is spotless. I run filters over all my intakes. Using Ceramique, my load temperatures are under 40C. It is definitely not a heat problem.
 
First thing I would do is check the BIOS, make sure all the settings in there aren't wrong. If that don't help, then I don't know what else you could do. Hope you get your comp going smoothe again, though. :)
 
In my opinion your motherboard is starting show it's age. Called it premature greying. Probably a memory controller problem or such. I would try another cpu for a few days and if the problem presists it's time to replace that motherboard. You seem to have checked for most of the other problems.

Buy you a cheap processor off the forum and then resale it after testing if your problem still presist. It might be worth your hassles for a cheap fix. After reading your sig, this might be your best bet.

I fought a simlar problem last year with a P3 550. I replaced everything including the CPU. The problem presisted even aftera cpu change. Once I shit canned the mobo, problem solved for me. This was my work computer and two other sytems were built at the same time and have the exact same problem. They both still have the reboot problem and will until the boards are replaced.
 
If it's a bad memory controller, wouldn't that cause errors in Memtest86+? That test comes out clean.
 
with my P4 2.4 (400fsb) i could over clock it too 155 for at least testing... when i first bought it over a year ago, now no matter how hard i try, i cannot get it to 150 at all..... Heat was also never an issue. I dont know if this is the same as you, but the chip or motherboard can no longer support that speed.
 
I have the same backbone (CPU, Memory, and Motherboard) you have, but only recently did I upgrade to 1GB of RAM. When I upgraded the RAM is when I noticed the same problem, granting it was about a year old at the time. So I'd be willing to suggest that it may be the CPU or Motherboard. In both cases it would be leads within the device that are losing connection with use. There are a number of ways to fix this without replacing the parts, for a short term solution.

I would suggest getting a BIOS that has the chipset voltage setting unlocked. This will allow you to get more Voltage across those older pathways that are failing randomly. I was just about to try this myself today, when I get off work. The thing to keep in mind is that this may increase stability for a short time, but the system probably won't last longer than 18 months without going back to stock speeds at some point, barring any other parts failing. Be extremely careful with OCing the chipset as ours is passively cooled... I think... Since I can't remember for sure, I better go clean out my case. :p
 
Optimus said:
When I upgraded the RAM is when I noticed the same problem, granting it was about a year old at the time. So I'd be willing to suggest that it may be the CPU or Motherboard.
Click to expand...

I just realized the leap of logic I made with this statement. In my experience, the RAM has tested fine in other systems. I, in fact, sent the first set back for replacement, thinking it was the problem, because memtest was showing about two bad locations in memory. The replacement showed the same issue, but in a different location. Also, if I clocked back down to 2.2 instead of the original 2.222, it would become stable again. That's 2Mhz (4Mhz DDR) difference which shouldn't have mattered to the RAM as I have personally seen this stuff hit 210 Mhz (420Mhz DDR). Thus, motherboard or CPU, since those are the only other 2 similarities between our rigs. ;)
 
1. Electromigration could occur making CPU less stable, needing more volts to maintain stability

2. Dust build up on components making them overheat faster, but more volts wouldnt help that (I don't think)

3. Motherboard components could be getting weaker with age. Crappy capacitors, etc... and more volts may make up for their weakness/ unclean signals.

Those are 3 suspects.
 
Probably a stupid question, but is the ram a matched pair for dual channel? I only asked because a while ago my comp was crashing all the time and I couldn't find out what was wrong with it. It ended up being the ram, although the chips were fine they just weren't matched for dual channel and that screwed things up
 
I'm having the same problem with my CPU. It's a P4 3.2C that I used to be able to get up to 3.72GHz @ 1.7V. Now it needs 1.76V to get into Windows and it will eventually crash. I have watercooling on the CPU and it barely gets above 40C under heavy load. It runs just fine at stock, but I want the spedd that I had before. Thinking about eBaying this guy and getting a new one.
 
ThomasE66 said:
They can 'age' in the sense that overclocking (more to the point, prolonged over-volting) can cause problems with electromigration and other effects that can shorten the useful life of a processor. It could also be a motherboard issue.
Click to expand...

The term in "Quantum Tunneling". Electrons fly off the transistor and smash into the surounding silicon. It damages it over time.
 
Alias said:
The term in "Quantum Tunneling". Electrons fly off the transistor and smash into the surounding silicon. It damages it over time.
Click to expand...

Damn that quantum improbability! Always messing up nice Newtonian determinism.
 
ThomasE66 said:
Damn that quantum improbability! Always messing up nice Newtonian determinism.
Click to expand...

LOL Stupid Laws Of Physics. Always biting me in the ass with its logic.

:p :D
 
The RAM is a matched pair, sent directly from Corsair's factory. I got a 2500+M, so I hope my problems are fixed.
 
mobile2500.jpg


AQYHA 0431

Now at 2.2GHz with 1.65V. Once I finish this Prime95 run, maybe I'll lower the voltage and see how it copes.
 
You must log in or register to reply here.
Back
Top