Search
Sign in with
Sign up | Sign in
Closed

Signs of a warn out cpu?

Tags:
Last response: in CPUs
Share

This might take some memory for those who long since moved away from the 478 socket days, but i have a question. My old Pentium 4 3.0GHz Prescott is starting to fail i think, i just want another opinion. Ever since one day my comp had some fits for a few days then everything seemed to be normal after then, except in some games they just lockup, other games such as Crysis 2 or Call of Duty Modern Warfare 2 just do a C2D. Its not the psu, ive hooked it up to a volt meter to make sure all the volts were fine, all readings were normal. i ran memtest on the ram and they all checked out after leaving it go for 8 hrs. HDDs are still fine, id be surprised if they were the culprit. Ive used ATiTool, Furmark to test the gpu temps and stability, and the gpu is rock solid even at its current max oc (810 core 990 mem on stock cooler) but the temps were pushing close to 90C in furmark but im getting a new cooler so they should drop down quite a bit (the gpu is also a Sapphire Radeon HD 3850 AGP 512MB GDDR3). I ran Intel Burn Test and it works the first time, but any time after that it wont go through, same thing with every Prime95 torture test, seems to be fine on the first attempt right after it boots up, but any attempt after results in a failure pretty quickly. It happens at stock speeds and OC speeds, no matter what the Vcore is set to, ive even tried 1.5v and it still wasnt running and still had crashes. CPU temps never go above 65C they should go down once i get the gpu cooler which will vent the hot air out instead of spew it all over inside the case, but the temp itself seems fine for me. Id like to know if i am correct that it appears to be a dying cpu or any other opinions/ideas/suggestions as to what this problem is or what the fix may be. (And yes ive tried clean formats and installs of windows aswell and still no luck)

Thanks, all help is appreciated!
Firedrake125

More about : signs warn cpu

Register or log in to remove

If you've got access to another Socket 478 rig, drop your CPU into that motherboard and test it again, to see if your mobo isn't the problem. However, I also think you should really find a different power supply that you know is good and hook it up to your rig for testing. Testing an idle power supply with a volt meter won't tell you anything about the PSU's behavior under load.
Anyway, my money is on an electrical failure, maybe capacitors on either the PSU or mobo. IMO, out of the three options, the PSU is still the most likely culprit, while the CPU the least.
CPUs Authority

Cpu's don't wear out, they either fire up or they don't.

I'm guessing it's probably overheating. The fan is probably worn out and not spinning like it used to or it's clogged with dust or both.

It could be anything and any suggestions would only be guesses Im afraid to say it's probably time to start thinking about saving for a new computer, I mean that's a single core cpu right?

PSU seems fine to me it was a brand new OCZ StealthXStream 600W one i got for a fairly nice price, its not even a year old yet and the rig doesnt run 24/7, ive used the volt meter on it while under load as well, probably wasnt the smartest, but nevertheless it told me the volts were in the range they need to be in, i used a flashlight and the caps in the PSU looked fine, all the mobo caps were good, also it wasnt the mobo because sometimes the whole rig itself wont post, also i just tested it with an "ancient" 400FSB 1.8GHz Pentium 4 Northwood, it was as solid as a rock at stock speeds, did a huge OC on it with an outrageous Vcore (1.65V! :D ) and almost got to 2.6GHz and it was still running stable, and cool 52C max load, so my odds are the cpu is dying, might of got fried that one day. Also all the other 478 rigs here cant run the 800FSB prescott(old things eh?)

update: yes its a single-core, but it does what i need it to do quite well. Handles Crysis 2 with no issues at all, quite impressive in my opinion. No its not overheating, 65C for a prescott is like running an air conditioner on it, ive seen them get to almost 80C before on stock speeds with no issues what so ever, but the 65C could be a bad sensor on my mobo idk, its not throttling at all. I checked it with throttle watch. Also i have been looking at a new comp, but im waiting for the day this thing just completely dies, or quits being able to handle the newer games.
Related ressources
Ask the community

Best solution

CPUs Master

cpu's can wear out, normaly when they have been oc' but have been stable.
what happens is they leak electrons and over a period of time the leakage gets bigger. you then have to keep adding more volts to get the cpu stable then 1 day it wont be able to take any more and either not start or go pop.
so if you want to know if it is starting to fail remove the oc if you have 1 running and set it back to default voltage. if it boots but its unstable then you can pretty much guarantee its on its way out.
CPUs Authority

HEXiT said:
cpu's can wear out, normaly when they have been oc' but have been stable.
what happens is they leak electrons and over a period of time the leakage gets bigger. you then have to keep adding more volts to get the cpu stable then 1 day it wont be able to take any more and either not start or go pop.
so if you want to know if it is starting to fail remove the oc if you have 1 running and set it back to default voltage. if it boots but its unstable then you can pretty much guarantee its on its way out.


Nope. We have never had an RMA cpu come through our shop that was worn out. It' either posts or it doesn't.
CPUs Expert

Well I would have suggested opening the case and looking at the CPU - if it is lying on its back and all the little pins wiggling around in death throes, then yes it is worn out and about to croak! :p 

Seriously, modern CPUs are designed to last for many years if not oc'd, overheated or over-volted (i.e., run at stock settings). Overvoltage can cause electromigration which is where the copper or aluminum signal and power traces become thinner and eventually fail due to the constituent atoms actually moving around. I still have a Pentium-150 that is oc'd to 166MHz, that works perfectly well since 1995 or so..
CPUs Master

geekapproved said:
Nope. We have never had an RMA cpu come through our shop that was worn out. It' either posts or it doesn't.
its unlikely you would as it takes time, the larger the oc the quicker it happens but your still looking at a quarter of the cpu's lifespan. b4 you will see any noticeable degradation in the cpus structurer. its a simple law of physics, where there is friction there will be wear and tear.

as you force more electrons down the lanes created in the silicon the lanes degrade leading to leakage as i said earlier. this happens at the atomic level. just because you haven't seen or heard of it doesn't mean that others haven't. intel discussed it in depth in a white paper when they moved from lo.K to HI.K materials for there dye shrink to 45NM.

like i say its very unlikely you would see an rma'd cpu because by the time it happens the cpu would be well out of warranty.

I agree and yes i have ran this poor Pentium 4 far past what it should be, in the winter i came close to maxing out my mobo with a 3.9oc all winter long and only a 3.4oc in the summer, mind you this is on stock cooling so i have no special cooler or liquid cooling at all. No the motherboard is still fine because i can pop in an old 1.8GHz Northwood and it runs fine. Which pretty much told me my cpu is burning up which Jay_83's suggestion about trying it in another rig reminded me that i had it, ive already ordered a new cpu for $19 and maybe i can push this one past 4GHz some time.
Register or log in to remove
 
Ask the community
Read discussions in other CPUs categories
!