Why is the ATX PSU designed to standby current?

My new Dells don't have power switches, just the annoying soft-power button on the front panel. Worse, they don't have reset buttons. Luckily, I use a UPS that *does* have a real power switch on the front.

John

Reply to
John Larkin
Loading thread data ...

There's a lot more than that. The typical PC also supports wake on modem (ring), wake on USB device interrupt (e.g., mouse moved), and wake on keyboard.

The whole standby power idea was specifically aimed at getting people to stop leaving their _entire_ computer running just so that they wouldn't have to wait for it to reboot. Standby typically uses more than an order of magnitude less power than active usage does, so it's a significant improvement.

---Joel

Reply to
Joel Kolstad

I would care to differ with the idea that wake on LAN is the only benefit the ATX power supply offers. Learn how to configure your computer to use suspend-to-ram, and try it out for awhile. Once you've gotten a taste for it you will never want to go back.

Unfortunately not all cheap motherboards and other hardware is fully compatible yet with suspend to ram, but hopefully your system is. When you suspend to ram all of the contents of your memory remain intact, but all of the power hungry equipment of your system turn off (monitor, drives, fans, processor, etc.). Power consumption drops dramatically, but your motherboard continuously refreshes the ram contents so they aren't lost.

As a result your computer shuts down in a matter of a couple of seconds, and boots up in a matter of maybe around five or so seconds. This represents a dramatic improvement over shutting down and booting up normally. Another very major advantage is that none of your applications need to be closed. When you turn your computer back on everything will be just as it was when you turned off your computer, all applications fully open and ready to use. This is a very valuable feature, well worth the extra power consumption (at least for me). But as others have indicated, if you still don't find any of the advantages of ATX useful, you are free to switch off your system fully using a mechanical power switch.

No need to lobby govt agencies to move technologically backwards back to using AT powersupplies all around. The ATX powersupply was designed with an importance of the environment in mind.

Reply to
Fritz Schlunder

There is this new invention called an "outlet strip" that allows you to cut the power to your monitor, computer, printer, etc. all with a single switch. Some of them even have built-in surge suppressors and interference filters.

Funny how "environmentalist" always seems to equal "he who calls for increased government power and decreased liberty"...

Look here for some eye-opening information:

formatting link

formatting link

Reply to
Guy Macon

Unless Windump has crashed so hard it's taken out the BIOS too. Then you have to hit reset before the power button will work. If it has a reset. I just cycle the big rocker on the UPS.

Something like 3% of the electricity ganerated in the US goes to power PCs left on all the time. They're left on mostly because Winmuck takes so long to boot. They use so much power because it takes a 100-amp Pentium chip to execute the 80 million lines of code required to print a 1-page document in less than 10 minutes.

John

Reply to
John Larkin

Unfortunately, not all OSs handle this properly. Windows, for example.

When you

Why doesn't Windows suspend to disk? Just copy all of ram to disk and

*fully* shut down, zero power. Restart would take about 2 seconds; spin up the disk, restore RAM, run.

John

Reply to
John Larkin

"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...

Windows does suspend to disk; it's called 'hibernation.' Unfortunately, it is noticeably slower to re-start from than 'standby,' adding about 10 seconds. Since re-start is being done by very 'simple minded' software, I think part of the delay is due to having to read in something approaching a gigabyte on a modern PC using the 'bombproof' but slow data transfer modes (e.g., programmed I/O for the hard drive).

I realize that, yeah, in theory it should take 'about 2 seconds,' but when you really start thinking about all the different hardware a modern OS has to support and how complicated the device driver stack to support all that is, Windows actually does quite respectably. In fact, Microsoft has people whose job it is to try to optimize Windows' start-up time; Windows XP even schedules periodic 'startup optimizations' whereby it watches how long various modules take to load when your PC starts and then changes their placement on disk and the load order to minimize the boot delay.

FWIW, Linux typically takes much longer to boot than Windows and I don't believe it has any automated 'boot optimizer' yet. Its support for standby and hibernation also seems a little less reliable than Windows, although I'd grant that -- at least around the 1998 time frame -- there was a lot of flaky hardware that no OS seemed to be able to get to suspend/resume properly. I haven't seen any PC built within the past three years though with such problems.

---Joel Kolstad

Reply to
Joel Kolstad

Uhh... I use suspend to ram with Windows 2000, works just fine. I've seen it work just fine with Windows XP as well. When it doesn't work, it is hard to be certain where to place the blame, since suspend to ram requires both hardware and software compatibility. However, in my experience all new versions of Windows support suspend to ram just fine, but sometimes specific hardware and specific applications sometimes have some trouble with it.

Windows 2000 and surely other versions of Windows also support suspend to disk. Windows 2000 calls it "hibernate." You can access the settings for it (and suspend to ram, and other power saving features) from the icon labeled "Power Options" from the control panel. If I recall correctly even in the days of Windows 95 (or at least 98) the laptop version of those OSes normally also supported suspend to disk. Suspend to disk really isn't as good as suspend to ram though for a desktop computer. It is much slower, especially if you have say 1GB of RAM. Even with a decent hard drive capable of reading at 30MB/sec., that would take over half a minute to load the contents back into RAM (or shut down for that matter). On the plus side though you don't lose all the data in your unclosed applications if the power goes out like with suspend to ram.

Reply to
Fritz Schlunder

Laying on the power switch to four seconds will shut down an ATX supply (at least that's what the spec says). It's just a slower reset. ;-)

Like everything there is a trade-off point. Cycling a PeeCee isn't all that hard on it. I generally power them on once a day. Once on they stay on. Though the reason I do so is more boot time than reliability/cost/power.

--
  Keith
Reply to
Keith Williams

I wish I has an OS that didn't need a gig of ram.

John

Reply to
John Larkin

Hmmm, I remembered that the "four second spec" was a PS hardware thing. I'll have to go find the spec again. I've never had a system crash so hard it didn't work, though I've had 'em lock up so hard I couldn't get them powered back on without cycling the AC (hardware fault, outputs crow-bared).

I've heard that statistic, but haven't seen it well documented.

100A across the *surface* of a chip is pretty amazing, in itself. Though they're not 100A because Win takes forever to boot or a printer is printing at .002 pps. They're at 100A because they can.
--
  Keith
Reply to
Keith Williams

It does, but it takes more than 2 seconds (takes perhaps 30 seconds on this laptop). ...longer to restore. There is more to hibernation than copying RAM to disk.

--
  Keith
Reply to
Keith Williams

It's significantly more complicated than that. Every single piece of hardware in the PC has to have its 'state' initialized to whatever it was before the PC was put into standby, or at least initialized to some 'reasonable' values. I.e., your graphics cards, network card, USB peripherals, etc. Windows 'walks' the device driver stack and tells them that the system is about to go to standby; the drivers then save their device's state to memory, the registry, etc.

What Windows actually does is to ask drivers if they're ready to go into standby or not. If they say 'yes,' Windows will then say, 'OK, let's do standby!' at which point the driver MUST shut down _even if new operations were started inbetween the time when Windows queried the device and when the standby command actually happens_. This makes for some rather tricky programming at times...

This is all especially fun if your device is one that supports waking up the system, such as a USB modem.

---Joel

Reply to
Joel Kolstad

I am mystified as to why the ATX PSU was designed so that it draws standby current even while the PC is off. I am a bit of an environmentalist and I find it rather lax of govt agencies to allow this blatent waste of what collectively is a whole stack of CO2 emssions.

The only advantage of the current (!!) ATX power supply design is the wake on LAN feature.

Alan Liefting

Reply to
Alan Liefting

How will it do wake-on-anything without drawing power? You could turn it "Off" if you didn't want to draw any power. And why is it the government's problem that you can't be bothered reaching for the power switch?

Ken

Reply to
Ken Taylor

My copy of Windows (Windows 2000 Server) has a suspend to disk feature. It doesn't work on my motherboard - probably something to do with the quad proceesors or the dual hotswap power supplies - but it's there.

Reply to
Guy Macon

If you live in a place that needs heating most of the year, does it really matter? ;)

-Lasse

Reply to
Lasse Langwadt Christensen

I wish I had disk drives which could restore 1GB of ram in 2 seconds.

Reply to
nospam

It does. It's called hibernate mode, and the problem is that much of the RAM contents have to be written to disk. For example, I have 1G of ram, and when I command a hibernate it usually has to save several hundred MB of memory to the hard drive, perhaps even 500MB or so, and this takes time. And it has to restore this at wakeup.

If you have a fast 10MB/sec drive, it takes 50s to restore 500MB. It might even need 100s when writing, for a read-after-write check? There may also be performance hits from disk fragmentation issues.

I'm not sure, but it seems Windows also wastes more time attempting to optimize the amount of RAM stored by requesting program action. There appear to be some timeouts while it waits for responses that don't come. And there are similar excess delays upon resume.

--
 Thanks,
    - Win
Reply to
Winfield Hill

Most of the time 3/4 of it is used to speed up the disk drives.

Reply to
nospam

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.