Good DC related thread...

[Spacehead]

... going on in Highly Technical.
Link

Is constant 100% CPU utilization good for your computer?
Interesting read, i wish i knew more about CPU's.

 

[BurntKooshie]

The thing to realize is that although the HLT command "pauses" the CPU -- it also causes voltage spikes, (if anyone here is an audiophile, they may hve noticed this.... sometimes it actually interfeares with sound quality). I guess the question, as one of the posters said, is which is worse: the constant use, or the spikes....

If you have poor cooling, that obviously makes teh constant-on a poor decision.

The final verdict? I don't know, but it's not a clear-cut answer.

 

[ProviaFan]

I'd upgrade before anything bad from running my CPU @ 100% utilization all the time could happen, anyway.

 

[Assimilator1]

Interesting point in there ,but as others have pointed out supertools arguments are flawed.

You can't compare a mechincal device like an engine ,which has many moving parts ,to an electrical device ,an engine suffers far more wear than a cpu does from electromigration

How often have you heard of a cpu dieing from electromigration(ignoring o/ced or overheated cpus) compared to an engine dieing?
Its rare for a cpu to peg out in that fashion.

He's also missed the point that running the cpu 100% via a DC project is making good use of the cpu

 

[RaySun2Be]

We've had this discussion a few times before, and IIRC, the main consensus was that running a PC (and the CPU) at 100% constantly would help minimize "wear and tear" (material fatique due to heat/cool expansion/contraction) but that the constant voltage would eventually cause problems. But that the CPU would probably be replaced long before the CPU died as we were talking years.

I posted a link to a thread at Overclockers, discussing the thermal cycle (thermal delta they called it) as being one of the main causes of CPU failure.

I wonder if one of the AT subscribers could do a search in the DC archives on the subject and find some of our discussions on the matter.



Thanks for the link.

 

[BurntKooshie]

How often have you heard of a cpu dieing from electromigration

According to sun, their UltraSparc III has a failure rate of about one in 83,000 due to electromigration degredation of the chip. So it DOES happen, even in non-overclocked chips (now, this probably speaks a bit more of Sun's.....although they're "high-quality", I honestly haven't heard any similar numbers from any other designer/manufacturer on this subject....although maybe everyone else is worse, I somehow doubt that though.....anyway, now I'm rambling....)

EDIT: apparently it's not copmlete failure.... the chip just degrades to the point where the FPU is that part with the 1 in 83,000 change of producing erroneous results.

 

[Assimilator1]

Hey BK ,I did say it was rare not unheard of
Also I was referring to personal experience not manufacturers figures ,though of course thats a very useful source!

 

[Insidious]

Didn't read the entire thread (yet)

But my training and experience with electronic components says that thermal cycles are the enemy assuming all else is correct. Voltage 'spikes' are bad because they cause localized thermal cycles (expansion and contraction of semiconductor materials) on the microscopic level).

The danger in continuous running is fouling of your cooling components (fans and heat transfer surfaces) if the system is not cleaned well.

My belief (15 years as electronic design engineer and 6 years as elec. tech. before that) has come to let me believe... your system will last the longest if it is on all the time assuming that it is kept clean and it has a stable power source.

-Sid

(constant loading of a DC also helps keep temperatures near constant.... which is a good thing if you are considering system lifetime)



(For your monitor..... hours of operation tend to cause aging... you wouldn't want it on 24/7 On, but blanked protects it as best it can be without turning it off all together)