Does keeping a program running on your computer, that utilizes 90-100% of the processor good for the computer?

[dxkj]

I have heard it argued before that it helps keep the computer running well. Can anyone explain that to me?


Thanks

 

[SuperTool]

Originally posted by: dxkj
I have heard it argued before that it helps keep the computer running well. Can anyone explain that to me?
Thanks

Does driving your car around at the redline when you don't need to go anywhere keep it running well?
It's a waste of electricity, and every time a computer circuit switches, the flow of electrons moves some atoms in the wires, so eventually, it could cause an open circuit.
While it's not very likely in normal operation for normal usage life of a CPU, it's a negative effect, and I fail to see what the positive effect of running a high load all the time is.

 

[CTho9305]

Originally posted by: SuperTool

Originally posted by: dxkj
I have heard it argued before that it helps keep the computer running well. Can anyone explain that to me?
Thanks

the flow of electrons moves some atoms in the wires, so eventually, it could cause an open circuit.

This is probably the major issue, assuming electricity is free. And it eventually WILL cause an open circuit, IIRC. Fans also wear out eventually.

 

[dxkj]

Seeing how most computers last only 3-5 years in use anyway normal wear and tear on a cpu is to be expected.


I have noticed that my system crashes less now that I have seti@home running, and Im basically trying to figure out why. Honestly Im ignorant to the reason why. Memory is cleared and reassigned more often? Stability is obtained from it being constantly used rather than having to start stop start stop?

Anyway its something I noticed, and Im not sure why and if its just happentstance, but Ive had others voice the opinion before so I thought I'd get some clarification if possible.

 

[CTho9305]

Originally posted by: dxkj
Seeing how most computers last only 3-5 years in use anyway normal wear and tear on a cpu is to be expected.

What kind of computers do you use? Emachines? My 486 from 1991 (?) was working great as a server until I threw it out a few months ago, Same for the p200mmx from 1996. Machines only die when they're used like AT members use them (constantly playing with hardware, etc.). If you use a computer intensively, and don't mess with the hardware, it will last a LONG time.

I have noticed that my system crashes less now that I have seti@home running, and Im basically trying to figure out why. Honestly Im ignorant to the reason why. Memory is cleared and reassigned more often? Stability is obtained from it being constantly used rather than having to start stop start stop?

Anyway its something I noticed, and Im not sure why and if its just happentstance, but Ive had others voice the opinion before so I thought I'd get some clarification if possible.

I don't know about that. Possibly something like the placebo effect... you remember crashes when not running SETI, but not those when running it? Maybe your usage pattern changed and you don't use the apps that do crash the system as much (e.g. changed the games you play).

It might be worth making a list of when every crash occurs and whether or not SETI was running at the time to see if it really does crash less.

 

[Shalmanese]

Yes, computers may physically last a long time but how many people have a practical use for a >300Mhz machine running 24/7? Again, only AT people who use them as servers.

The things that are most likely to die out over extended use are fans and monitors. As an aside, if you leave your computing running at 100% 24/7, what you have effectively is a ~150W space heater. Thus, if your in a cold climate, having a room full of computers means your getting free heating out of it.

 

[TJ69]

I don't think the comparison to revving a car engine to red line is fair. Maybe revving to peak horsepower or torque RPM range is a better comparison. Sure the engine will suffer normal wear and tear but nothing out of the ordinary. It's doing what its designed to do and may even out live you with normal maintenance (if its made in japan ).

A cpu has no moving parts. It requires no maintenance. So why would it fail? Well it may overheat and fry. You may crack it installing a heatsink. A voltage surge may cause a short. "the flow of electrons moves some atoms in the wires." Maybe over time the silicon its made of becomes brittle and the likely hood of it cracking will increase until finally the tension of the clips crushes the chip.

If anything, I think it heat that causes a chip to fail or to fail quicker. But with adequate cooling, even at 90% load it can easily outlast the use you have for it today.

 

[JHutch]

The upside to running at 100% is that the temperature stays relatively constant. Granted its relatively high, but it is constant. That means less expanding/contracting on parts on the system board. Whether this translates to better stability or longer/short life? Probably too little either way to matter. And this is from someone who runs DNet (currently) 24/7 on all my machines.

JHutch

 

[calbear2000]

Originally posted by: SuperTool

Originally posted by: dxkj
I have heard it argued before that it helps keep the computer running well. Can anyone explain that to me?
Thanks

Does driving your car around at the redline when you don't need to go anywhere keep it running well?
It's a waste of electricity, and every time a computer circuit switches, the flow of electrons moves some atoms in the wires, so eventually, it could cause an open circuit.
While it's not very likely in normal operation for normal usage life of a CPU, it's a negative effect, and I fail to see what the positive effect of running a high load all the time is.

Yep, all circuits are susceptible to electromigration or self-heat failure. RV checks (reliability verification) are done especially on custom circuits to see how long before EM or SH occurs given normal activity.

 

[mundane]

Not completely in line with the question posed ... but doesn't Winows [2K] have an Idle Process running anyways? Even if it effectively does nothing (add two meaningless numbers, jump to the next instruction, or a blank instruction, or ...), the CPU must still go through the fetch/decode/execute cycle.

Granted, this process would not need access to memory or IO devices, so there would be no added wear and tear to those elements. But the issue at here is the CPU itself. If the CPU is in constant [relative] use anyways ... what's the difference?

Of course, to me this means that performance/stability/lifetime on the CPU won't be compromised by running, say, SETI@home. As with the others, I do not see how always running an expensive process constantly would be beneficial to the system.

I'm not terribly familar with the x86 processor or how the Windows designers implemented this - I'm curious about it. Any gurus on hand?

-Josh

 

[CTho9305]

Well, what you said is true, but what if your engine was only idling instead of revving at 7000 rpm? Obviously less of a problem. Or, maybe only fire one cylinder instead of all of them.

 

[Webthug]

I'd agree with diegoalcatraz the CPU must still do something is it is not doing anything that an application has asked it to do. It can not just stop doing things. Therefore it must have a system idle process. I am lead to believe this is an on chip process, something that is built onto the chip. The principal difference between running SETI@home, folding@home, etc is that the calculations that the processor is using are usefull. The heat thing is associated with the CPU being required to use more if it's on chip hardware than when it is using the system idle process, thus there will still be atom movement with the associated flow of electrons in the core part of the ALU will still ocour with or without the CPU running at max.

 

[CTho9305]

Originally posted by: Webthug
I'd agree with diegoalcatraz the CPU must still do something is it is not doing anything that an application has asked it to do. It can not just stop doing things. Therefore it must have a system idle process. I am lead to believe this is an on chip process, something that is built onto the chip. The principal difference between running SETI@home, folding@home, etc is that the calculations that the processor is using are usefull. The heat thing is associated with the CPU being required to use more if it's on chip hardware than when it is using the system idle process, thus there will still be atom movement with the associated flow of electrons in the core part of the ALU will still ocour with or without the CPU running at max.

Again, HLT instructions stop a lot of the processor. Besides, I would tend to think that electromigration and other such "wear and tear" are most significant in the processors critical path. If a non-critical path gets slowed down due to thinning interconnects, it is OK since you have some margin for error. If a critical path gets affected, you're going to have to underclock. Then, the question is just: "Is the critical path in the parts of the chip used for HLT and the other idle instructions, or in things like the FPU which are used by SETI but not an idle process"

 

[RaySun2Be]

Here's an interesting discussion, along with some formulas.

Overclockers Discussion on CPU damage

One thing they discuss is the thermal delta, which was the consensus at one time of being the biggest cause of CPU wear and tear. The expansion and contraction of the materials as they heated up, and cooled down, instead of it being on and a fairly constant tempature at all times.

One thing to account for is the spike/surge at start up. Hard drive platters having to spin up, fans spinning up to speed.

I guess I'm lucky but I've never lost a CPU to running overclocked and/or running 100% CPU 24x7. I've had hard drives fail, fans fail, Power Supplies fail, a stick or two of memory go bad, and one motherboard stop working. But the CPUs still all work.

And yes, some people still use their Cely300s, PII450s, out there. If you aren't into gaming, and just surf the net, they can be sufficient in use. I think it's a lot of ATers that feel the need to go out and get the latest and greatest, gotta have the fastest CPU, Mem, and VID.

I'll see if I can't dig up some of our old thread discussions on this topic, as they used to come up from time to time in the DC forum, due to the large number of PCs running 24x7.

 

[Webthug]

i'd disagree with some of the stuff on the website, because i have a PII that I purchased whn they first came to australia, and that has run 24/7 practically ever since i purchased it. That means that it has been in constant opperation for arround 40,000 hours and it still works fine. Perhaps i'm lucky but i think that there is deffinatly some validity in leaving the comp running 24/7. The only thinks that have been changed are the ram (32MB just was not cutting the cake) increasing it to 1024MB and I outgrew a 3GB HDD and have replaced it with a 20GB, never them less the orrigional ram and HDD are still in operational condition.

 

[pillage2001]

Originally posted by: CTho9305

Originally posted by: Webthug
I'd agree with diegoalcatraz the CPU must still do something is it is not doing anything that an application has asked it to do. It can not just stop doing things. Therefore it must have a system idle process. I am lead to believe this is an on chip process, something that is built onto the chip. The principal difference between running SETI@home, folding@home, etc is that the calculations that the processor is using are usefull. The heat thing is associated with the CPU being required to use more if it's on chip hardware than when it is using the system idle process, thus there will still be atom movement with the associated flow of electrons in the core part of the ALU will still ocour with or without the CPU running at max.

Again, HLT instructions stop a lot of the processor. Besides, I would tend to think that electromigration and other such "wear and tear" are most significant in the processors critical path. If a non-critical path gets slowed down due to thinning interconnects, it is OK since you have some margin for error. If a critical path gets affected, you're going to have to underclock. Then, the question is just: "Is the critical path in the parts of the chip used for HLT and the other idle instructions, or in things like the FPU which are used by SETI but not an idle process"

Isin't it just like running NOP for an infinite cycle until a software/hardware interrupt occurs???

 

[JWMiddleton]

Not completely in line with the question posed ... but doesn't Winows [2K] have an Idle Process running anyways? Even if it effectively does nothing (add two meaningless numbers, jump to the next instruction, or a blank instruction, or ...), the CPU must still go through the fetch/decode/execute cycle.

The value of Idle Process = 100-sum(running processes) It only runs at update to calcute the value; thus, Idel Process is just that. What a waste of CPU power!

 

[dmcowen674]

Originally posted by: JHutch
The upside to running at 100% is that the temperature stays relatively constant. Granted its relatively high, but it is constant. That means less expanding/contracting on parts on the system board. Whether this translates to better stability or longer/short life? Probably too little either way to matter. And this is from someone who runs DNet (currently) 24/7 on all my machines.

JHutch

I'll chime in too.

I have seen tons of equipment die from constant on and off.

I have seen a good share of equipment that is left on.

The main failures I have are mainly due to physical parts like fans seizing and pregnant Caps.
I have actually seen more fans die from the constant on and off equipment that the 24/7 equipment. Most of the time the next time a machine is turned on the fan does not rotate and you get thermal runaway resulting in either a Power Supply failure or Motherboard failure, most of the time a CPU will shut down before the heat makes it fatal.

There is no doubt the equipment left on fairs better.

Also, I have never seen a CPU die from 100% duty cycle exhaustion due to running processes like DNET or SETI.

Also the heat topics are an absolute joke as well. As long as you do not have thermal runaway (as mentioned above) and within defined manufactured specs, heat will not kill.

 

[ScottMac]

Aside from a laptop or two, all my home systems are running 24x7 (about a dozen at any given time)... in most cases, for years (I've got a Novell server that's been up for a little better than two years).

All run some flavor of Seti@Home.

If the hardware is solid and the software is good, it's not a problem.

(...and I just went over 21,100 WU in ~2.7 years)

FWIW

Scott

 

[Eskimo]

CPUs are tested and verified to run for a certain useful lifetime assuming 100% utilization (or as close as one can come to that). There will be no difference in useful lifetime between keeping a CPU fully loaded and only utilizizing it off and on assuming it is adequetly cooled in both situations.