Running 100% bad for the life of your CPU?

[memo]

I've read some articles that running your CPU at 100% can lessen the overall life of it, is this true? I wouldn't mind cruching out some units for a good cause, like curing cancer but i don't want it to damage, or lessen the life of the proc. are there any articles out there that state otherwise?

 

[Coquito]

A constant state is better then frequent start ups & shut downs.

 

[memo]

Originally posted by: Coquito
A constant state is better then frequent start ups & shut downs.

i leave my computer on all the time, the question was though, whether running your CPU performance at 100% all the time damages it.

 

[ssvegeta1010]

Also, a question for you all, would running a laptop at 56C about 6 hours a day shorten life? (Chip maxes out at 85C)

 

[ecvs85]

I've read somewhere that the safe temperature is around 20C below the maximum temp.

I've a question too, would running an athlon 64 at 59C 24/7 shorten it's life fast? (maxed at 70C)

 

[BlackMountainCow]

I look at it this way: YES, 100% will shorten the life of you CPU. But the question is: HOW MUCH?

I've had comnputer runing 24/7 at full load as servers in my father's office that never failed once due to a CPU problem, and I'm talking about stuff like old 386/486/PentiumI & II stuff here. In my own apartment, a Pentium III/AthlonXP and Pentium 4 are running 24/7 100% load. I never ever had a CPU fail on me. I assume that 100% shortens the life of my CPUs but by the time it would really be a problem, they are so old that it doesn't matter anyway. Maybe some of those 386/486/Pentium would fail on me in the future, but I don't run them anymore. So, YES, 100% shortens your CPU life but it just doesn't matter.

:beer::beer:

 

[Cawchy87]

Overclocking will take more years off a cpu than leaving it on 100%. But yes, the components will de-generate faster.

 

[BadThad]

I find that hard to believe that using a CPU at 100% load would shorten it's life. CPU life is shortened by one thing (AFAIK), and that's electromigration. All silicon based chips undergo this process, it's accelerated by high voltages (currents). If the CPU is run designed voltage, the process is so slow it would take many, many years to be affected. Overclocking with higher voltage will increase the rate of electromigration, however, by the time it has the "effect of death" (assuming you don't use an insanely high voltage), the CPU would be obsolete. I'm talking about 5-10 years!

I've read a lot of threads over the years regarding this phenomena, including those posted by an Intel CPU design engineer who used to post on AT. Personally, I have no reason to believe that a 100% CPU load would shorten the CPU's life....someone prove me wrong, hehehehe.

 

[memo]

Originally posted by: BlackMountainCow
I look at it this way: YES, 100% will shorten the life of you CPU. But the question is: HOW MUCH?

I've had comnputer runing 24/7 at full load as servers in my father's office that never failed once due to a CPU problem, and I'm talking about stuff like old 386/486/PentiumI & II stuff here. In my own apartment, a Pentium III/AthlonXP and Pentium 4 are running 24/7 100% load. I never ever had a CPU fail on me. I assume that 100% shortens the life of my CPUs but by the time it would really be a problem, they are so old that it doesn't matter anyway. Maybe some of those 386/486/Pentium would fail on me in the future, but I don't run them anymore. So, YES, 100% shortens your CPU life but it just doesn't matter.

:beer::beer:

yeah that does make sense i guess. so sign me up! im going to be doing FaD, just started a WU(is that what its called?) and signed up on teamanandtech.com, how do i go about linking the two so my stats can count for TA? what the procedure to join?

also, what are the sizes of these WUs? like a couple KB? just curious to see how much i would need to upload/download.

 

[ShotgunSteven]

Your CPU will be outdated long before it is in any danger of dying. I ran SETI@Home on my old Athlon 1.4 GHz for years and it never had a problem.

 

[ecvs85]

Don't forget that the CPU life depends on the condition of the CPU itself too. If you got a crappy CPU sure it's life is much shorter. I'm hoping mine would fall into the average group or better still the "tough" group

 

[GLeeM]

TeAm number is 2039 for FaD.

CPUs are designed to run at their rated speed. If they didn't have to consider that the CPU might be run at 100% they could raise the rated speed way up.

EDIT: If a CPU goes it is usually something else that goes and takes the CPU with it.

 

[memo]

my job has been stuck at 5.79% for about 45 minutes now, it actually went up to 5.79 in about 6 minutes but it seems like its stopped. my task manager says im still running at 100% though.... :-/

 

[GLeeM]

The percent done is not very accurate. When there is a hit it seems to take longer to get past that part.

Some finish in 30 minutes - no hits. I had one take over three days to finish - with 115 hits.

 

[mrwizer]

About the laptops, I have been running my P4 1.8 M with DC projects for 2 years now. No Issues, although it does get hot! And I have heard mentions that heat can cause the life of the battery to be lessened, but I have not experienced this myself. Plus, my battery only has a life of 400 discharges. So some may have reached the life before they notice heat related issues.

 

[xComputerManiacx]

BadThad is right about electromigration being a real problem. As current flows through the silicon layers in the chip (at the transister level), the gap between the positive and the negative channel gradually widens. Over a very long period of time (could be years if it's not completely pure silicon and/or not very well designed, or it can be decades), it can grow too wide for current to continue flowing, so 'dead chip'. It's late so I don't remember much else, but I can pull out my course notes if someone's interested. *has a B of Sc degree in Computer Engineering*

 

[mondobyte]

There are a lot of myths .... but there is a kernel of truth to this one ...

If we operate our CPU at 90% or more of maxium operating temperature 24x7 - then the CPU will probably fail long before it becomes too slow to be useful irrespective of the CPU load.

If we operate our CPU at a constantly varying temperature substantially lower than the maximum operating temperature, for example halfway between ambient (25C) and maximum operating with the temperature varying +/- 10C from that baseline then the CPU is still likely to fail because of the constant thermal stresses. This is particularly the case if the CPU is frequently power cycled or if the load is abruptly varied. Power cycling and extreme CPU load changes create thermal stress and this thermal stress is more likely to result in failure.

If we operate our If we operate our CPU 24x7 at a constant temperature (or very slowly varying temperature) substantially lower than the maximum operating temperature, for example halfway between ambient (25C) and maximum operating temperature with the temperature varying +/- 2C then the CPU is likely to outlast a CPU that is rarely used but power cycled with each use.

All the studies that I have read suggest that thermal stress is a reason for failure by many orders of magnitude compared with electron migration.

The only CPU's that I have seen fail in service are those where the CPU Heatsink/Fan physically separated from the CPU or the fan failed. In both cases the CPU temperature exceeded the maximum operating temperature by a significant factor. I have actually seen CPU's get so hot that they desoldered the socket off of the motherboard or charred the motherboard.

I have seen CPU's run at greater than the rated maximum operating temperature 24x7 for long periods of time (> 1 year) and not fail until they were power cycled and thermally cycled (allowed to cool down). Typically, they did not fail in service.

If you purchase a computer and plug it in and run it 24x7 at any constant CPU load (provided no component runs at a temperature higher than 85-90% of max) then it probably will not fail in service.

If you take an identical computer and power cycle it four times daily and subject it to highly variable loads, mean time to failure is significantly less than 2 years on some component.

If you take a third computer and power cycle it very infrequently (> 3 months between usage), I suspect the computer will fail more often than either of the other two computers.

I've run many computers 24x7 since the 90's and I always run some artificial load to keep the temperature relatively constant.

I find it interesting that most mainframe operating systems have tasks that run when the cpu is idle to maintain the internal temperature of the mainframe. The goal is to keep the entire mainframe at a very constant temperature.

mondo

 

[DaFinn]

Running them 0% is bad for your life... go for it

 

[dmcowen674]

Originally posted by: mondobyte

If you take a third computer and power cycle it very infrequently (> 3 months between usage), I suspect the computer will fail more often than either of the other two computers.

mondo

Good post Mondo.

The bottom line is that "In-Rush Current", the surge of electrons that occurs on start-up is a lot more "damaging" to electronics than remaining on over the lifetime of an electronic device regardless of the "load" level.

 

[ecvs85]

Thanks mondo! Your entire post was what I'm looking for this whole time. Since my CPU temp is in the range of 86% of max temp, I went out, bought and install the zalman CNPS7000B-Cu and my temp dropped 7C to 53C under load (of course that's without the 200hour burn in for the arctic silver 5 yet), but now I'm not worrying about my CPU anymore

 

[Ken g6]

I started DC on a Cyrix (for the uninitiated, Cyrixes ran very hot, and were fairly unreliable). Its fan failed at least three different times, crashing it, but it recovered each time. It took five years before it started having problems, and even then I couldn't diagnose any hardware malfunction, and thought it might be Windows 95.

 

[ssvegeta1010]

Mondo,
My CPU is a Mobile P4, so according to you, i should try to keep my cpu either clocked all the way up, or all the way down, instead of throttling with SpeedStep?
I guess 56-60C is far enough from 85C. Also this computer is actually only on for small amounts, but i guess ill try to keep it on longer, if 24/7 is actually better.
Thanks!

 

[AgaBoogaBoo]

You'll get HD issues before you do CPU issues very likely

 

[mondobyte]

I agree ... components with motors and moving parts will usually fail before CPU's

 

[ssvegeta1010]

ok, then that is fine, i wont worry.