Are you sure about that? I used to play with those old Seagate drives quite a bit, and I could swear I remember hearing the heads snap into their parked position the moment I cut power. I always figured they had some sort of storage cap that dumped its charge into the voice coil when power was cut.
I think the head park was a big selling point for Seagate once. I remember at one point they claimed they were building an artificial reef out of all the notorious IBM PC OEM drives that had been replaced by Seagates.
Come to think of it, waiting for the drive to spin down would be very bad. It would cause the head to drag across the platter, virtually guaranteeing damage.
I remember it being a huge selling point about voice coil actuators on the 251-1. I was selling computer parts wholesale at the time, so the flurry of calls we got was huge.
My roommate worked in the tech department at the same company and couldn’t stop going on about how cool the vca was. They assembled all the computers that our buyers bought pre-built, and their rule was to wait until it was powered down. As you say, you can hear sounds that tell you when it’s done, but you want everything spun down all the way before you pick up the case and stick it in a box for shipping, or move it to another desk, or start slapping in new components or whatever.
Anyhow, that was back in the 80s, but I’d still say it’s always a good idea to let everything “spin down,” as in let the fans visibly stop spinning, and the other sounds you hear inside the computer stop. It doesn’t take 10 seconds, more like 4 or 5. I just wouldn’t want to reapply power to something mechanical that has capacitors and still has motion going on (hard drives and fans, for instance).
I wait because a) I’m a geezer who remembers when it counted and b) I worry about the mechanical components, and stress on them.
Power off buttons on computers these days don’t turn off power, but go into a shutdown sequence.
I wouldn’t worry about left over bits in the memory. The power on sequence for modern microprocessors is very complicated - I’ve sat through hours long meetings on this. You clearly make no assumptions about the state of any register. I’d assume that the first thing you do is to invalidate all the caches. I’ve never had to deal with a computer boot procedure, but I suspect it is orders of magnitude more complicated still.
I can certainly see the security issue of stuff left in static ram after power down, but that is not going to hurt the system.
As for initialization, I just did a man malloc and it still says the block of memory is uninitialized. Unless programmers are a lot better now then when I was writing big programs, I doubt very much there are no bugs left.
Well you can say it like some components will experience transients and some won’t.
Btw, I realize that the components with stable power will stay stable and don’t actually need to be reset. Unless, that is, their state got corrupted and you’re restarting the computer just to fix it. That’s when things get scary because the trusty reset doesn’t help.
This piqued my interest! Did a bit of searching… Turns out, the “minutes” figure only applies if your computer’s in the deep freeze. Fifteen seconds is more typical.
The attack depends on encryption software holding its key in RAM during use. The OS supposedly blocks unauthorised access to RAM. But if you interrupt the power, you remove the OS while leaving data in RAM. If you chill your stolen laptop, and then pull the battery a moment, you can boot it from an external device. Your purpose-written, small memory occupation, extract-the-RAM OS can then read the RAM directly. Neat.
It’s probably only useful for stolen machines with careless owners, though. In fact, good encryption software should overwrite any key it hasn’t used for a few minutes. So you’d have to steal the computer while the encryption software is running, with the key already entered and in use. Which probably means you don’t need the key…
When a computer has to restart, i.e. automatically shut down and reboot, there’s no apparent waiting period going on, so I can’t see why I would need one for a manual restart.
What OS are you using? I usually have to wait at least a few seconds after requesting a shut down before the machine actually shuts down. In Windows sometimes it refuses to shut down for hours (well, it seems like it) while it installs updates.
Anyway, that’s different. The only reason to actually cycle the power on a computer is if something’s gone wrong, which usually means there’s corrupted memory somewhere. In the old days, and maybe even yet for some adapters and whatnot, the corrupted data could still be there if you merely flipped the power switch off and on.
The chance of any residual RAM contents causing a problem was always incredibly small, but waiting a couple of seconds cuts that chance from small to nil. That’s a small price to pay to eliminate a variable when you’re debugging something.
The other part of this is the ten seconds. When you’re telling someone (for example, over the phone) to pause between turning the switch off and on, they want to know how long. The fact is, nobody knows exactly how long is enough (because you’re trying to prevent something unpredictable). The ten seconds is just a nice round number that’s almost certainly long enough.
Face it. If you tell somebody to wait ten seconds, you’re likely to get a random delay anyway.
If everything is working properly, you don’t need to turn the power off. It’s when some circuit is not working properly that you may need to turn the power off, wait a bit, and turn it back on. Ideally, there would be a hardware reset button and/or signal that would put everything in a known and working state.
The reality is that sometimes you have to cycle the power to fix a problem. This can sometimes happen when the equipment take a power hit. The power supply voltages drop low enough to screw everything up, but not low enough to put everything in a known state. Sometimes design engineers don’t think about, or test for, that sort of problem.
So… from all the replies, even the ones I don’t understand the tech jargon completely, it seems it’s not really necessary anymore, but it’s not a bad idea to do it either.
I asked originally because my old CPU had so many bugs and ran windows 98, it would normally freeze up every 15-30 minutes by the end of it’s life. So I’d have to hold the power button in to manually turn it off. But I’d still do the “wait 10 seconds thing” each time.
Good point, but I think the trick is that Windows is locked because the person who left their computer unattended was trying to be careful. Of course if the comp is on and there’s no password it’s all more trivial. I don’t think most encryption software overwrites anything, although that will hopefully change with time.