CPU Apparently not working

[CitrixEngineer]

I say "Apparently", because I've never seen a CPU simply die without showing at least some form of flakiness beforehand: especially when that CPU has been fine for a little over a year (just out of warranty!).

RIG:
ABIT BX-6 mobo
256MB Kingston PC-100 SDRAM
3dFx Voodoo 3 3000 16MB AGP
SoundBlaster Live!
Hauppauge WinTV PCI Nicam
3Com 3C509TX 10/100 NIC
SoundBlaster AWE 32 8MB
WinModem
Quantum Fireball 4GB
Seagate 3.1GB
Maxtor 17GB
Pioneer 105 16x DVD
Sony 4x4x24 CD-RW

Scenario:

I've been running this computer as my main PC for nearly 3 years. The software build is MS Office 2000, IE5.5, DX8, Magix Music Maker 5, Steinberg WaveLab, Steinberg Cubase 3, Microsoft Age of Kings and Civilisation II Test of Time. I don't run shareware or install anything else on this machine because I use it to write music on (and play the occasional game!). I have another 6 PCs on my home network to play with software on.

Once a month I spray ice around the inside of the case to drive out the dust, and there are 3 fans: the usual one, and a pair located front and back - one reverse wired - to provide constant airflow.

Last January I upgraded the processor to the current Intel PIII 450. The processor is held in place by a huge heatsink with a bolt at each corner that goes through the mobo. Although this is an ABIT board, I would rather buy a new processor than overclock this machine.

Two weeks ago I installed Windows ME. On Sunday evening, I got the message "The CPU has changed or is unworkable".

I can get into the BIOS, but I cannot set the CPU to anything that makes the computer start. Even factory defaults will not let me get past the message.

Does anyone have any ideas (apart from "Buy a new proc&quot?

Thanks in advance.

 

[CitrixEngineer]

I've seen it now. It's official. Processors really do just peg out and die after 1 year and 2 weeks of near-continuous operation.

I couldn't even get into the BIOS last night. I swapped the processor for one that I knew was working - no problems at all.

Strange that the 486 I bought in 1992 is still going, though. Maybe it's a case of "They don't make them like they used to?"

Technical discussion or ideas still welcome!

 

[edfair]

Traces erode like the filament of an incandescent light bulb. At some point they open up. If an address line then you are toast. This sounds more like a failure in microcode addressing. So you are now partial holder of the left end of the bell curve of lifespan of P3-450s. Unfortunately, it lasted just long enough for you to help Intel's bottom line when a replacement is purchased. Most of what I've seen of sudden death is within the first couple of days. Ed Fair
efair@atlnet.com

Any advice I give is my best judgement based on my interpretation of the facts you supply.

Help increase my knowledge by providing some feedback, good or bad, on any advice I have given.

 

[CitrixEngineer]

Yes, that's what I usually see. A couple of weeks at most. After that, most CPUs go on for ever, it seems! Sadly, the shop I bought it from stuck (legitimately but unhelpfully) to their 1 year warranty policy.

Still, the new 733MHz processor takes advantage of my motherboards' ability to run the bus at 133Mhz very nicely.


Thanks for your response, Ed. Another one for experience.

 

[kd4yuw]

Just read the thread and I have a question. Did you "burn in" the new processor? That is leave the pc running constantly for a couple of weeks. I had always heard that PCs would last forever if they made it past the first two weeks of constant operation but I never knew how much faith to put in that credo.

Humbly
David

 

[CitrixEngineer]

That's probably a good idea, David - but do you know anyone who does this? I certainly don't!

The two week rule of thumb is one I've heard - and this is certainly the first time I've seen a processor just keel over like this.

I've had one other very good theory put to me - a situation that might have arisen when I installed Windows ME:

When you switch your PC off, the fan(s) all stop operating. Obvious, right?

OK, what do the fans do? They cool the insides of the case down, and there's one important one strapped to the CPU that keeps that cool. No rocket science here.

If the fans aren't in operation, what happens to components that have just been shunting electrons about for a few hours/days/weeks?

They suddenly get very hot, since they are now allowed to release all that energy as heat in one go. This can be demonstrated with a basic thermometer - it's that significant a rise.

Now, if you restart these components over and over again, you are putting extra strain on all of them at the same time. The weakest link is the one with the narrowest path for the electrons to travel down.

Ed's ananlogy with a light bulb is very pertinent in this case. If you have millions of extremely fine filaments under this kind of strain, it's possible one might overheat and break. If it's the right (wrong!) one, then the whole CPU is toast.

Modern CPUs can cope with losing a few of these strands - all that will happen is that a particular circuit cannot be made, error checking will kick in, and the calculation may take a few nanoseconds longer.

The original processor from this machine is still ticking away merrily running my #2 PC, 3 years on. Since I use it to watch DVDs and as an integral part of my home hi-fi, I'm reluctant to take this one down.

Thanks for your reply.

 

[kd4yuw]

We (wife and I) have tried both leaving the PCs on all the time and shutting them off. At my last count I think I have 7 PCs here in various uses but only one is now left on 24/7. The others are generally turned off and on as needed. I have the main PC left on mainly because we check in quite a few times through the day to get e-mail or check a website so its faster for us to just leave the thing on. I did have a LAN setup and the main computer also handled the proxy server and dial-up duties also so we needed it to be on. Now that I think about it, the computers we have bought new and burned in have been almost maintenance free (aside from software problems nothing hardware related). The others do seem to get "sick" more often. Don't know if that's related to powering off and on but it's an interesting observation none the less.
Humbly
David