Slashdot Mirror
Five PC Power Myths Debunked
snydeq writes "Turning off PCs during periods of inactivity can save companies between $25 and $75 per PC per year, according to Energy Star, savings that can add up quickly for large organizations. Yet most organizations remain behind the times on PC power management, in large part due to common misperceptions about PC power, writes InfoWorld's Ted Samson, who outlines five PC power myths debunked in a recent report from Forrester, ranging from the energy savings of screen savers, to the energy draw of powering up, to the difficulties of issuing patches to systems in lower-power states."
this article was written by a self-aware PC who is tired of the human race's waste of time and energy.
My favorite line from TFA is the last one: "The Forrester report "How Much Monday are Your Idle PCs Wasting?" is available for $279." Please raise your hands if you know someone who would buy that!
Energy is kWh power is kW. "Energy at a rate" is power, and should be in kW not kWh.
One of our competitors trademarked the term "hypothesis". From now on, we will call them "boneheaded ideas".
In the winter I leave my computers on. I don't think I am "loosing" any energy that way since it's used to heat my house.
>>>"Turning off PCs during periods of inactivity can save companies between $25 and $75 per PC per year"
How am I supposed to download last night's episodes of Smallville and Supernatural if I have my PC turned off during the day? Jeez. Insensitive clod. ;-)
FOX NEWS.com should be BANNED from television and internet. Have the Congress take it over and give us Truespeak.
If you spend 10 mins per day turning you pc on and setting up your work environment, and 5 mins closing everything, the cost of your time spent on this task will negate $25 saved ten times.
May Peace Prevail On Earth
Myth No. 1 really hurts to read. I'm not sure there is a single instance there where the units of power and energy are used correctly.
Can you be Even More Awesome?!
Turning off PCs during periods of inactivity can save companies a substantial sum. In fact, Energy Star estimates organizations can save from $25 to $75 per PC per year with PC power management
Lets assume each PC has a user who is paid at least $25000 per year. We can clearly see the savings on the cost of that employee and thier PC setup caused by this are negligable.
he Forrester report does acknowledge that end-users have very little patience for downtime. However, it suggests that "potential user complaints can be mitigated by communicating the positive financial and environmental benefits of PC power management."
Complaints or not the company is paying for any user downtime.
note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
. . . as if millions of Folding@Home and Seti@Home clients suddenly cried out in terror and were suddenly silenced.
All through college I left my PC on 24/7, however now that I'm paying the bill I have thermal throttling and the other new power-saving standards all turned on, and I turn everything off (router, modem and all) entirely when I'm not using it. It's odd the way people look at it; at work some users say "Well I never leave it on at night because I know that it makes the computer die quicker" and some people say "Well I never turn it off because I want it to last longer." I think the truth is that modern hardware really can handle both philosophies and it's just a matter of convenience vs. power costs at this point.
"potential user complaints can be mitigated by communicating the positive financial and environmental benefits of PC power management."
Now that just plain hilarious.
Sites like Blackle suggest that a black screen saves energy. May have been true for CRT displays, but modern TFT Displays always have the backlight on, even on a black screen.
Learn how to save $25 to $75 by purchasing the $279 dollar report that the article is hawking. No thanks. This article has no business even being on Slashdot. It isn't news, it is an advert.
Tequila: It's not just for breakfast anymore!
1.42kw for the computer to run overnight has a cost of around 10 cents to the company.
Waiting 5 minutes for your PC to boot at the federal minimum wage of $6.55 per hour has a cost of around 55 cents to the company.
It costs the company at least 5 times as much to have you boot your PC in the morning as it does to let it run overnight.
Moderating "-1, Disagree" is simple censorship. Have the guts to post your opinion.
I agree that you can save power with low-power (standby) modes on your PCs.
However, as a network admin as a mid-sized company, I also have seen loads of frustrations where PCs (both laptops and desktops) don't come out of power save mode cleanly, requiring a reboot. Wake-on-lan is also a great concept, but also pretty buggy (again...in my limited experience trying to implement it). We also have issues where our client systems are using network applications with license pools (e.g. database applications or CAD packages). When a user leaves one of these applications open, then the PC goes into power save mode...it really freaks out when it comes back out of power save mode since the license server thought the system had released the license, but the client still thinks it has a licens in use. This situation usually results in the need to reboot, which frustrates the users to no end.
I set all of our PCs here to lock and send only the monitor into low-power mode after 20 minutes or so. Then we don't have the problems with coming out of power save mode and having locked up or frozen applications (especially the aforementioned network applictions), but still save a good bit of power by allowing the monitor to be turned off automatically.
Anyone have any idea what percentage of power is used by the monitor versus the PC itself? I don't have a clue, but I'd bet it's a pretty good percentage. There's also probably a big difference between CRT monitors and LCD monitors...again, my gut feeling, but I can't cite any numbers.
Later,
JS
S3 is such a nice feature. My wife never powers down her computer all the way any more, just suspends it to RAM, in seconds, and the boot up is just as fast. That said, the last 2 motherboards I've used, while technically support S3, are unable to suspend without immediately waking up. I've done my homework on it and no matter what I do, it won't stay suspended (unplugged all USB and network cables, only had a monitor and ps/2 keyboard and it still doesn't suspend). Does anyone know of any websites that have a list of motherboards that properly implement S3 mode?
WTF is with this stupid meme?
Of the four instances in which watts were referenced (directly or in compound units), three are completely boneheadedly wrong:
They should be:
You *can't* call it a typo when they are perfectly backward in three out of four incidents. And you can't call it "They just got it backward..." when they got it right once. You must conclude, therefore, that they have almost no grasp whatsoever of units.
FTA: Modern computers are designed to handle 40,000 on/off cycles before failure
With all the reboots required, that means I am limited to three Vista reinstalls?
"Myth No. 5: My PC users will not tolerate any downtime for power management.
The Forrester report does acknowledge that end-users have very little patience for downtime. However, it suggests that "potential user complaints can be mitigated by communicating the positive financial and environmental benefits of PC power management.""
I love this kind of response. It's pretty much ignoring the problem. PC users will not tolerate any downtime for power management even if you "educate" them. This is trying to wave the problems away and it won't work.
Knowledge is power. Knowledge shared is power lost.
Any modern PC can S3 suspend.
S3 suspend cuts power use by 95% and the PC resumes *INSTANTLY*.
I can S3 suspend my laptop and have it run off the battery for over a week - open it up and I am back where I left off in about 2-3 seconds.
There is no argument against having an IT policy MANDATING S3 suspend. Hell you can even automate it to do it by default every day at 6 PM unless the PC is in use (easily checked by screensaver APIs).
Or you could use a cron job and wake on lan to shut them down at night and start them up in the morning without affecting the worker drones at all.
When our name is on the back of your car, we're behind you all the way!
Then there is the issue of starting up for the day. Shutdown can happen automatically, but startup should be initiated by the user. Sometimes it does take several minutes to connect to online volumes or for MS to do whatever it does. I have seen a couple machines take a very long time to boot. Again, I think hibernate is a good compromise, but there must be hooks in the system to allow virus updates and other patches.
All this means that all applications must be closed in case a automatic update occurs, something I almost never do on my machines. I put them to sleep, but my apps are open. On my MS Windows machine, this every once in while means I have to start all over again loading apps.
"She's a scientist and a lesbian. She's not going to let it slide." Orphan Black
I don't use suspend usually. The main reason being that it kills stateful TCP connections.
ssh is the big one here for me. Things are made a good bit easier by using GNU screen, but I still need to re-establish a bunch of ssh sessions, many of which are dual-factor authenticated.
Another is the inability of people to send me IMs when the system is suspended.
Etc.
I would have liked to know how much more the computer uses when it is booting up (or closing down). I might turn the computer off when going for lunch, but with the data presented in the short article, I cannot determine how long you have to be away from the pc to make it worth to shut down the pc. It boots in under 1 minute, so the time I lose by booting is negligible (I have to boot my head as well after lunch, and focus on coffee - that takes at least a minute as well).
I have never had a PC or a Laptop which was able to reliably "Suspend" or "UnSuspend" Never in my life.
Not with Windows or several Linux Distros. I would say at least 25% of the time the machine will not return and must be rebooted anyway.
From my experience with some corporations, the way it works is more like:
1. The left hand doesn't know, and doesn't want to know what the right is doing. If your department can save $10 bucks, but it costs everyone else 10 million in workarounds and lost productivity, who cares? You're the greatest anyway.
2. Any attempts to rein in waste and such effects, just introduces one more layer who'll get their bonus for making you buy a tool that costs $10 less, but where you spend 100,000 more in salaries to do the same job. Occasionally it introduces a masked form of corruption too: they get more bonus for buying a $1000 pencil at 50% discount, than a normal one at 5% discount. In the former case they "saved" $500 per pencil. They're that great.
3. Don't underestimate interdepartment power games. Making you curse and waste more effort for implementing my hare-brained cost-cutting schemes, is the gretest achievement some people can get. It's me having power over you. For some people it's a powerful drug.
4. Theatre. Being seen as doing something beats doing the right thing. You can see that at all levels and in all domains: security theatre, cost-saving theatre, etc. Being seen as being teh great green saviour can beat actually saving money.
5. In that vein, beware the new boss who just has to piss on everything to mark his new territory. The higher level, the more dangerous. These guys _have_ to show that they changed something. It shows vision, leadership, etc. So he'll cheerfully make an actual loss, just so he can put a good leadership and vision theatre.
6. There's a whole caste of people across the pyramid whose goal in life is to not rock the boat and not be responsible for anything. It's better to comply with a dumb rule (even one that wasn't supposed to apply to your situation or domain) than to have anything be your personal decision, and responsibility if it fails. Applying someone else's rule is like having a papal indulgence: whatever goes wrong, you're not the one who'll be punished for it. These fine guys and gals would mindlessly enforce even turning off the computers _during_ work hours, if that's what the rules say.
7. Don't underestimate the effect of rewarding failure. E.g., see the thing about "saving" money by buying a disproportionately _more_ expensive thing. E.g., in some places, keeping the people under you from doing their job can mean needing to hire more people, and if you get enough of them you get a promotion. E.g., being the guy who dumbly applies rules without thinking, cam actually get one a promotion or at the very least it's often enough to not get demoted or phased out.
So, yes, I've seen places where they paid consultants in the range of thousands per hour, but would rather pay those to twiddle their thumbs for a quarter of an hour while a baroque configuration starts, than "waste" cents on leaving that computer idle over night.
A polar bear is a cartesian bear after a coordinate transform.
I keep wanting to just ignore such errors, calling the writer an idiot, but often the random units cause the science to be indecipherable. In this case you can figure out what they are getting at.
Any modern PC can S3 suspend.
S3 suspend cuts power use by 95% and the PC resumes *INSTANTLY*.
I can S3 suspend my laptop and have it run off the battery for over a week - open it up and I am back where I left off in about 2-3 seconds.
There is no argument against having an IT policy MANDATING S3 suspend. Hell you can even automate it to do it by default every day at 6 PM unless the PC is in use (easily checked by screensaver APIs).
I still have issues suspending/waking computers. Generally it works fine... But sometimes you run into odd issues.
One client we support has a piece of software that hates waking from suspend. Pitches a huge fit. All sorts of errors.
And I still have problems with some computers/OSes that really should handle S3 just fine simply choking on it. Won't resume reliably or whatever.
The real problem I have with power saving options is rolling out the settings consistently across multiple computers. Last time I checked (and it has been a little while since I checked, so I could be wrong) there was no way to push out power settings with a GPO. Sure, you can set screensaver options... Turn off the monitor or something... But that doesn't get you a suspended computer. You can set options on the individual computer, in their motherboard settings... But that isn't easy to update/change across a network. You can throw together a pile of scripts to shut down machines...maybe try to use wake-on-LAN to power them back up in the morning...
I'm not saying it can't be done. And I'm not going to say that you can't save any power by doing it. But there doesn't seem to be a simple way of managing these settings across a network yet. It still seems that power management is a hacked-together feature that was tacked on after the fact.
I'd love to be able to push out a group policy that made all the computers on my network suspend after an hour idle.
"Work is the curse of the drinking classes." -Oscar Wilde
Try a Mac. One of the things that initially impressed me most about my old iBook G4 was that sleep actually worked. (I have a Thinkpad X60 tablet now and while sleep mostly works, I never know whether it's going to suspend to RAM or to disk.)
"[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz
It makes perfect sense. I just read it to mean that power usage increases every hour by 89 watts, and that by the time it's been left on overnight for 16 hours, the power usage of an average desktop has increased to 1.42kW. (If you think I'm joking, do the math -- that can't be a coincidence ;-))
This kind of growth of power usage means it is extremely important to turn off your PC every night, otherwise by a month later the power consumption will be approximately 64kW, which will probably result in it melting through your desktop unless you have very good cooling.
Not only that, but "watts per hour" doesn't actually make any sense at all. Unless we're talking about something that is gradually consuming more and more power as time passes.
I would always leave my bedroom PC on in case I needed to FTP into it and get some data when I was away from home. Since then, though, I've learned that it's easy to power up my PC from anywhere in the world. Then I VNC into it, do stuff and shut it down.
It really feels like I'm living in the future! Actually, my computer is set to auto-hibernate when there's no activity for a while, and WOL can wake it from this as well. These days I also wake my computer from work before I go home, and set it to download the previous night's Colbert torrent, so that it's ready when I get home. Now I need some sort of a USB-switchable power strip so that I could control the power of my other appliances, like lights and audio system.
I totally agree, S3 suspend really hasn't matured to the point where it can be used without repercussions. Lots of software tend to crash when waking up from S3 suspend, or even S2 standby. Especially those god awful wireless network card drivers. And once they go down, your network card simply wont be active without a restart due to sudden jump in time. Too many things can go wrong on HAL. Even when using linux. Also, some hardware simply need the bios to re-initialize them, OS just wont do the trick and they stay at S3 even though the rest of the computer is back to active mode. I think we should make the "green" features functional before preaching about using them.
Where is the "Ignorant" mod tag?
This kind of growth of power usage means it is extremely important to turn off your PC every night, otherwise by a month later the power consumption will be approximately 64kW
64kW should be enough for anybody.
It's valid, ironic criticism of Wikipedia. All too often people -- who should know better -- quote Wikipedia for some proof of something here on Slashdot. Of all people, slashdotters should know that wikipedia is often an unreliable source of information, and that the link they've quoted can be easily changed after it's posted.
The "citation needed" is also an ironic criticism of the deletionist, book-burning nazi-pedants who appear to make up a significant core of the wikipedia admins. And yes, it IS possible to have it both ways -- a reliable source without the nazi-pedant zealotry. It just that this is not the direction wikipedia chooses to go in. Thus, it's the worst of both issues -- unreliable and exclusionist.
Long may this meme continue, or at least until there are significant changes in the way Wikipedia is run.
The linked article is nothing more than a poorly disguised plug for an expensive Forrestor research paper.
One thing I never see in any of these power-saving articles, even a recent one on how to save energy cooking your Thanksgiving turkey, is how the waste heat ties in to your HVAC system.
It makes a huge difference to your math depending on whether you need air conditioning, heating, or can passively cool with simple ventilation.
So, let's assume you're in an environment like mine (Toronto, Canada); 6 to 8 months of the year you need at least some domestic heat. (The radiators are usually started in October and shut down in April at my office.)
If you need to ADD HEAT to an office or house, then every Watt you save in electricity, you have to replace some other way. Now, given typical technologies, like an 80% AFUE furnace, it's about 30% cheaper to do that with the furnace, sure. But that means that $75/year savings is actually only 30% of that amount (because the furnace has to run more often). Or it's ZERO if you use resistive electric heat! (Currently oil heat is slightly more expensive to run than city gas here, but not as expensive as resistive electric.)
You can't just shut everything down over night and let the building cool off completely. It must be at least... well, it's 68 degF for residential, so let's assume commercial office space has to be similar. You could have a set-back thermostat let the temperature drop a bit, and then boost it back to normal before the workers get in, sure. And maybe, in a very well-insulated building, it would still be worthwhile turning off some of the machines.
If you need to COOL an office of house, it goes the other way. Using an EER 13 air conditioner as a reference, I worked out that 100 Watts of heat require an additional 42 Watts of cooling power to remove that heat from the air. So, in that case, your $75 GROWS by 42%. (And if you're in a warmer climate, it will grow by even more, as the same air conditioning plant will work less efficiently.)
But it's not that simple, either. We're talking about PCs left on overnight, yes? Well, the cooler it is outside, the more efficient an air conditioner works. So, actually, it costs more to remove the heat generated by the PCs DURING THE DAY, and much less over night. Especially if your system can switch to pure ventilation when the outside temperature drops below the inside thermostat set-points. (Whether or not that can happen depends on locale; in Toronto, it's only a couple or three weeks of the year where it's warmer at night than you want inside.)
About the only time these articles calculations make sense is, when you can just open the windows and have the interior space at the right temperature.
Unit errors are generally a sign in technical fields that a report hasn't been well thought out. No engineer proofreading this would have missed such a blatant error which means that an engineer didn't proofread it.
If an engineer did not proofread it, an engineer did not likely do it. Therefore, the content of the article was likely done by an incompetent hack and charging $279 for the report is a way of hiding the fact that it was written by a hack.
Great Scott! Wait til the PC hits 88mph, then you're going to see some serious shit!
You see? You see? Your stupid minds! Stupid! Stupid!
260 work days per year
5 minutes of employee time wasted per day
1300 minutes wasted per year, or 21.6 hours!
Even if you take a minimum wage employee at $6.55 per hour, that's a cost of $141.92 (larger for higher paid employees) per employee for powering off the PC. Sure seems to me a smarter investment to leave it on!
Yeah, you can figure out what they are getting at. But do you trust those figures to be correct when they clearly have no idea what watts and watt-hours actually are?
If you mod me Overrated, you are admitting that you have no penis.
It's really disappointing that other vendors can't seem to get this right.
Indeed. Especially considering that this feature has been around for EVER. My father's first powerbook was a 520, bought in 1994, which had the same sleep capabilities as todays macs, and I believe it has been around longer then that (possibly already with the original Mac Portable).
Almost true. Mostly true? Or used to be true. Or, in a perfect utopia, this isn't true anymore.
There's a tool from Energystar called EZ GPO which lets you install an power managment agent on the client, and manage it using an administrative template. In my experience, it works pretty well. It's a bit weird though: for some configurations, the tool doesn't use the win API, and has broken in the past with Windows Updates.
AFAIK, Windows 2008, or a Vista workstation on a 2k3 domain can be used to manage power savings on XP if the client-side extensions are installed.
Also, some expensive tools like LANDesk support power policies. Not ideal for most people ($$$$), but if you're already using "enterprise" management tools, worth checking out.
" "Modern computers are designed to handle 40,000 on/off cycles before failure, and you're not likely to approach that number during the average computer's five to seven year life span."
Too bad all major HD manufactures claim 10,000 power cycles, and many power saving settings will turn off a HD w/o doing anything else. Which means you may have many more than 1 HD power cycle per computer power cycle.
"some studies indicate it would require on/off cycling every five minutes to harm the hard drive."
over how much time, because if you did this continuously, you would kill a harddrive in less than 35 days since you would have eat'n all 10,000 average power cycles.
Server hardware and consumer PC hardware are different. Server hardware doesn't traditionally get powered down (hell, many servers have dual power supplies in case one fails), so the hardware manufacturers don't test this as much.
But the bigger issue is software. If a desktop PC doesn't boot cleanly, it's a bummer, and one employee will lose some productivity, but ultimately it's not a big deal. If server processes don't come up, the stakes are a lot higher.
When you press the power button on that server, are confident that everything will come up cleanly? How confidant are you? Would you bet your job on it? If you keep your payroll server down for the two weeks it is idle, try to bring it up to run payroll, discover that it won't come up, would you be comfortable with having your boss on line 1 asking where the fuck his paycheck is, your boss's boss on line 2 wondering the same thing, the Department of Labor on like 3 demanding to know why they just got 1000 complaints filed against you and how you are going to resolve them right fucking now?
Sure, the server will probably boot cleanly and doom will probably not strike. But when it does, the cost is going to exceed, many times over, the savings you'll ever see on your electric bill by powering-down a few servers.
They don't grade fathers, but if your daughter's a stripper, you fucked up. --Chris Rock
Many people think that kWh is "kiloWatt PER hour", and confuse watts for energy and watts-times-hour as power. It looks to them as a division rather than a multiplication. They wouldn't distinguish kWh from kW/h. This guy is, therefore, three out of four times right :-)
I'm glad I'm not the only who noticed this. Their blatant misuse of units is irritating and shows their lack of knowledge. But they are right in saying a computer booting up doesn't use 16 hours worth of power. I work in a cube farm and most people don't turn their computers off at night, some people even leave CPU intensive screen savers on without any monitor power down feature, so their CPU is thrashing all night long and their monitor is showing everything for nobody to see.
Yep, if your PCs are managed enough that you can get them all to boot up before most people will use them and shut down after most people have gone home - that's a good solution. And, the A/C factor might add a cost factor of 2-3x in the summer (with a somewhat balancing reduction in heating costs in the winter, depending on where you are.)
Devil's advocate will now raise the (dismissed in the article summary as outdated) concern of heat cycling your chips and any marginal solder joints they may have, as well as starting and stopping your hard drive's spindle and any wear effects on the bearings. I have had many 8-10 year no-failures computers that chug along nicely being run 24-7, they get retired because they are hopelessly out of date. I have also had many computers with significant (motherboard, hard drive) failures after 3-5 years of daily power cycling...
Many is a word, that only leaves you guessing
guess about a thing, you really ought to know...
$75 dollars per year? that is pathetic.
If it takes an employee 1 minute to power off (including properly shutting down all apps and saving shit), and 4 minutes to power up and launch all those apps again.
That's at least 16 hours a year. Best case scenario that employee costs $20/hour so that's $320/year...
Most of the time employees cost more like $40/hour all in.
So the moral of the story is keep your PCs running all year to save lots of money.
Wait, were they supposed to be debunking or spreading myths?
I've traditionally left my machines running, to avoid thermal stress from power-cycling and mechanical wear on parts from spinning up from a dead stop. I've found the big savings comes from two things:
If I need more power savings, I might spin down the hard drives. But modern drives don't use that much power just to keep the platters spinning, most of their power consumption's driving the heads. Simply retracting the heads and not moving them lowers the drive's power consumption by a fairly big percentage, and that'll happen automatically when the system isn't accessing the disk. None of this requires any fancy sleep or hibernate or suspend magic.
I have noticed one thing, though. My Linux systems go idle fairly cleanly. Nothing's happening, minimal CPU time gets used (mainly the regular cron process waking up to check whether there's anything to run, then going back to sleep) and the hard drive stays completely idle. Windows, OTOH, keeps pinging the hard drive every 5 seconds or so even when completely idle. It's not much, just enough to make the HDD light flicker, but I don't see that with Linux. It makes me wonder how much of the "You need to put your system to sleep!" hype is simply because Windows doesn't know how to idle properly?
What I'd really like to see is power settings based on the time of day...
7:55a - wake the PC up, I'm coming into the office
8a-5p - High performance mode. Display off after 30min. Suspend/Hibernate - never
5p - Power savings mode. Display off after 10min, Suspend/Hibernate - 30min.
3a - wake up for downloading updates, virus scan, backup
4a - back to power savings mode.
XP and Vista don't have this. Not sure about Linux, OSX.
Are there any utility apps that can do this?
Can this be done via network admin?
Just an opinion (based on experience as a user of PCs since before suspend-to-memory existed) - suspend-to-memory works, really well, except when it doesn't.
With each new OS (Win95, 2K, XP, OS X, Vista), I have given suspend-to-memory a fair shot at working for me, I turn it on, I use it, it's usually less than a month before it screws up and causes me to have to hard-reset the machine to get back to a working state. This is on notebooks and desktops from all types of manufacturers, Dell to Apple to Gateway to Homebuilt, even a few "Corporate managed" fleet machines.
I have no idea why they (OS and hardware vendors) can't make it work right, I just know as a user that I get more consistent and predictable performance from my PCs if I don't use it.