Peter Smithson wrote:

I've done some reasearch on the web and I'm now pretty sure that the increased CPU heat only applies to laptops with AMD or Intel chips. Something to do with a HLT instruction which has been around for some time.  I got this info from the following links -

Good question that has an easy simple answer and a much longer detailed answer...

Assuming just PC 'desktop' and 'laptop' computers, I'll try a brief stab at an answer:


Computer electronics are usually designed for 100% utilisation for 100% of the time. Running your CPU at 100% utilisation is fine.

Newer computers have "green" energy-saving features that are controlled by "APM" or "ACPM" that can switch off unused devices. Hence, if your system is idle, it can shut down into an unresponsive 'sleep' state. Some trigger event then switches things back on to bring the system back to life. This is very good for laptops to conserve battery power, or just to reduce generating heat.

A HLT ("Halt") instruction can be used on the CPU which literally stops all further CPU operation until some trigger event. This reduces the very large CPU supply current down to just trivial leakage current while in the halt state. (However, the supply capacitors get hammered by violent transients when entering or leaving the CPU halt state. This may reduce their life slightly.)

Conversely, running CPU tasks that heavily utilise the FPU increases slightly the current consumed over just running integer routines.


So... for running s@h:

The CPU heat output will be slightly higher than for an idle but otherwise fully powered up system. If your heatsink and fan are ok, then this is no problem (1 or 2 deg C temperature rise). If you have a marginal near heat-death system, s@h will be the first application showing problems or causing problems.

Keeping s@h active means that you have an active system, so no or limited 'power saving' by ACPM.

On laptops that are a design compromise due to their size and power, this may reduce their expected life /slightly/. On desktop PCs, there should be no concern whatsoever.


Further points:

The greatest system stress on computers is during power up and power down. You could argue that frequent off-on cycling by the 'power-saving' stuff costs more in wasted time and reduced component life than what is saved. (In other words, switch your work PC on in the morning, and switch off just once when you leave work for home.)

Constant CPU load means constant heating which means constant temperatures which means /minimum/ mechanical stress to give the greatest lifespan.


There's other factors, negative and positive also.


In summary:

If your machine is switched on doing something, then its spare CPU cycles may as well get usefully used (for no adverse impact to the system or you). The extra power consumed is small. 'Wear and tear' effects due to s@h are negligible amongst all the other effects.

One proviso: Don't have your disks spin-up and spin-down many times an hour!

Note that your computer will become obsolete long before it fails!


Thus is just my opinion. And I have all my machines run at 100% CPU utilisation while they are powered up. Never burnt out a CPU yet. (Meanwhile, lots of peripherals have 'come and gone' (:-|)

Happy crunchin',
Martin