Should you turn your computer off or leave it on?

The article's writer is unclear on the concepts of power and energy.

6EQUJ5: .... which leads you to conclude .... what exactly?

I agree with 4 out the 5 points. But the disagreement is literally a system killer: Frequent power cycling desktops really does kill machines sooner than not power cycling them. I've observed over the last several years multiple instances of machines dying on a power cycle with failed hard drives, CPU, memory and motherboards.

And the best example I have is six essentially identical desktop machines we bought about five years ago. Of the six, three were run essentially 'non-stop' - never being powered down except for things like an OS or hardware upgrade (new video card or the like) or dust cleaning. The others were frequently powered up and down as needed.

Guess which three of the six are still working. The others *all* developed serious hardware problems causing them to be taken out of service.

The only problem I've had with the continuously run machines is the need to replace the CPU and case fans as they wore out and started 'growling'. CPU/case fans are cheap and easily replaced.

At any rate, it's a good arguement for the South Pole. The fuel we burn to keep us powered is much more important (especially coming into winter and our flights are behind schedule on our fuel resupply) than the maintenance of pcs. thx for the info!

I always power down (and switch off at the mains), ever since a colleague told me how many house-fire scenes his (firefighter) wife has visited which were caused by faulty computer power transformers.

The fan in my old laptop went kaput yesterday, and I am a notorious user of the "sleep" option, powering up periodically for 5-10 minutes or so, every couple of hours, all day long. Can't help but think I killed it...

The real killer is thermal cycling. When you power up the PC, everything warms up and expands. When you shut it down, everything cools down and contracts. Repeated thermal cycling causes solder joints and pads to develop microscopic cracks. Eventually, they fail.

Disregarding the fact that the author of this article doesn't seem to understand the difference between a kW and a kWh, let's take his numbers and assume that a PC consumes 1.42 kWh during the 16 hours a day that it might otherwise not be consuming power. Obviously, electricity rates vary, but at 10 cents/kWh, that's 14.2 cents worth of electricity. The cost of one PC repair, not even considering lost data and other associated costs, makes that 14.2 cents per day a trivial cost.

I used to manage a corporate PC fleet of about 1,000 systems. Systems that were power cycled every day failed much more often than those that were left on 24X7. And, contrary to what the author of this article says, modern systems are more likely to fail than systems of five or ten years ago. Manufacturers are still having problems with lead-free solders, which in addition to causing shorts with microscopic tin feathers also are much less able to withstand thermal cycling.

Consider incandescent light bulbs. When was the last time you were sitting there reading and one burned out? Probably never. Incandescent bulbs fail when they're turned on or turned off, and it's because of thermal cycling. PCs are subject to the same problem.

Question: How does this relate, then, to "sleep" and "hibernation". Hibernation is the same as "off" I assume, but I'm not so sure about sleep.

Can't help but think I killed it...

Dear Ana, you can't blame yourself like this. Its time had come, that's all. There was nothing you could do to prevent it. You're only human. And your laptop ... it's, well, only a laptop.

Its time had probably come ten months ago, when I replaced the hard drive, not without considerable effort given the blind install and pins falling out... I've been told that a fan replacement is much more complicated, even impossible, so it seems I will have to use that new line of credit for more than dental work. HP won't even accept it for trade-in now, it's so old. I do like their new machines, though (HDX 16t series)...
Cheaper to buy now or after xmas, i wonder?

The "sleep" option seems to mean different things for different computers. The desktop here doesn't power-down at all (that I can tell) while sleeping.

"it's, well, only a laptop."

Heretic.

We have 500+ computers in our building, and I had to win some major arguments to even get a monitor shutoff time set on the screen savers. Because "users would be confused by the need to shake the mouse or hit spacebar to wake up the monitor". And you know what? They were right; the most common service call we get, more than a year later, is that someone sits down at the computer and there's no picture, and they don't even touch the mouse before calling us. We've put up signs, sent out email notices, to no avail.

Then about a third of users refuse to ever log off because it takes too long to log on. (Guess which OS we use?)

We've also tried (at my insistence) to implement wake-on-lan so we could power down the machines at night. Couldn't get it working in our network environment for some reason.

The problem is that energy is distributed and paid for at the University level, so the colleges have no incentive. A friend of mine came up with a brilliant idea that will never be implemented: put each college on an energy budget based on last year's usage. And here's the important part: let them keep half the money saved by their energy savings, for up to ten years over the original baseline usage. There'd be a sudden blossoming of energy awareness on campus.

Personally... I leave my desktop running continually, with an aggressively-timed screensaver that also starts up a BOINC client. So when I'm not using the machine it's still put to good use, and I don't feel too bad about the power used because
a) I usually choose lowish-power components,
b) I undervolt where possible, and
c) our electricity comes from renewable sources.

My desktop is usually the most powerful PC in the house: the older machines get turned off when not in use as they're less useful for distributed computing.

Robert Bruce Thompson -

I am curious if that extends to laptops. I try not to start it for a momentary email check, but I usually leave it shut down when I'm not using it. I don't actually take it anywhere, because it's ancient (I also had to rebuild the power connection with plasticine, among other reconstruction issues).

Honestly, I probably won't leave it on, because some of the work I've done on it makes me a little nervous about potential fire issues. But as I am getting set to start back to school in January, I am probably going to pick up a new(ish) laptop that will spend most of it's time where this one is.

What if you just heated the computer up slowly in an oven before you started it.

DuWayne, one of our faculty members has a new Dell that can boot in two different modes. In Windows, it's sucky and slow and presumably uses a lot of power while wearing out the hard drive and running the fan overtime. But it also has a Linux mode that doesn't use the hard drive. It runs off a no-moving-parts internal flash drive, and starts up in, literally, a handful of seconds. I think HP may have a model that does something similar.

The result is that he doesn't mind shutting it down when he isn't using it, because he can bring it back to life very quickly when he needs it. He really likes it a lot. One of those might be a good solution to momentary email checks.

Yes, like Barry said, you can leave your computer on with the distributed computing client of your choice. This way, you avoid temperature cycling problems, and since your computer is put to good use, you are not actually wasting away power. Let's put it like this: if a scientist needed some computation power, s/he could a) buy computing equipment or b) use the already available through a DC project. True, specialized equipment consumes less energy/FLOPS, but think of the (environmental) cost of producing that equipment.

About fire risks, well, i don't think the actual computer catches on fire, probably much more often it's a bad power outlet or an extension cord. Those can be easily replaced, and should be.

Paladin -

My laptop really is a fairly high fire risk - technically I don't think it qualifies as a laptop any more. I don't have the ability to replace my power supply and even if I could, it is now a permanent part, thanks to the plasticine I used to remake the plugin to my computer. But it's not the power supply that concerns me the most, it's some of the borg reconstruction I have done to it.

I needed more USB ports and couldn't afford to buy a hub, so I basically made my own from parts cannibalized from a couple dead computers. I then hardwired the assembly in place. My monitor's mount was broken and I had to get inventive about reconnecting a torn connection and used the sides of a clementines crate to mount the monitor back onto the body of the computer. The pictures a little shaky, but it works.

As soon as my student loans come through, I'll be buying myself a brand new piece of shit laptop. But until they do, I make do with what I have. But I absolutely do not leave it on when I am not around, because it really is as likely as not to catch fire eventually. I know very little about computers and little more about micro-electronics (though I am very capable when it comes to wiring a house or repairing bad wiring in the home). What I do know is pretty much self taught. I put things together in a way that seems to make sense to me. Occasionally fire happens. (actually, that could be said about a lot of my hobbies, mainly when I was growing up - occasionally fire happens)

I think this entire discussion has to be different for laptops vs. non-laptops, especially regarding the laptops (like the early titaniums) that occasionally spontaneously burst into flames. We have an old titanium here that gets hot enough to cause a first degree burn. In your lap.

George -

I wish I could afford something fancy, but if you read my above comment, it should be clear that I'm on a tight budget. I am looking at a toshiba satellite for about four hundred bones. But it's ok, I don't really need to check my email when I run in for a minute. Really, I don't.

Well ok, so I do. But my son's computer is usually on (his is a nice one) and if not in use, running with a DC project through his school. It costs very little to run and I have a good friend who works system security for a major computer manufacturer and helps keep it safe (he also helped us get far more computer for my son than we could have otherwise afforded). But I don't like to use his computer other than those really quick, gotta check it moments, as I am really hoping it will get him to middle school....

Greg -

...especially regarding the laptops (like the early titaniums) that occasionally spontaneously burst into flames.

So in effect, your saying that reconstructing bits of it with balsa like wood is a bad idea? I think I have some aluminum flashing - I'll just take a gander for them in the morning and change things around a bit.

Thank you very much for this useful information.
Please keep on .
I am looking forward to read your next great article.

I think this entire discussion has to be different for laptops vs. non-laptops, especially regarding the laptops that occasionally spontaneously burst into flames. We have an old titanium here that gets hot enough to cause a first degree burn.