Slashdot Mirror
Power-Saving Web Pages: Real Or Myth?
An anonymous reader writes "Are dark webdesigns an energy saving alternative to a snow white Google? The theory is websites with black backgrounds save energy, based on the assumption that a monitor requires more power to display a white screen than black. Is this a blatant green washing ploy by Blackle.com, or an earnest energy saving tweak for a search tool we use every day? To find out, PCSTATS hooked up an Extech Power Analyzer to a 19" CRT and a 19" LCD and measured power draw — turns out there is a not insignificant difference ..."
Did anyone here actually believe this? The big power draw is from the backlight, which is still running even with black pixels.
If you can't convince them, convict them.
"turns out there is a not insignificant difference "
Double negatives are not not bad.
If you wait a few seconds, your watts will turn into joules.
buy an LCD (or LED) screen. That will save much more electricity than changing the colors you use on it. I can never figure out why so many energy saving tips focus on such small things (e.g., turn off the water when you brush your teeth) but ignore the big issues (like my neighbors who water all afternoon in 100 degree heat and have a stream of water running directly into the sewer).
Oh god. I was wondering why my screen randomly seems to increase/decrease in brightness.
I hate this feature. It makes me think someone slipped me some acid, and then I'm disappointed, because no, it's just bad attempts at saving power.
The idea is valid for all of the smartphones running OLED displays. OLED's take no power (or very little) to display a black pixel. It takes full power to display white.
"Are not dark webdesigns an energy unsaving alternative to a snow white Google? The theory is that websites with black backgrounds don't save energy, based not on the assumption that a monitor requires more power to display a white screen than black. Is not this not a earnest endeavor by Blackle.com, or a not earnest not green not washing not not not not not ploy by not Blackle.com? To find out, PCSTATS didnt't not hook up an Extech Power Analyzer to a 19" CRT and a 19" LCD and measured power draw — turns out there is a significant difference ..."
Mine would have been shot down for being too readable though.
Support the EFF and Creative Commons. The war is coming, and they're supporting you...
There is a not insignificant parsing complexity.
"Flyin' in just a sweet place,
Never been known to fail..."
This.
It is obvious that black is good for the earth and white is bad.
Why do you think we have climate change? Because of white, of course. No one has even heard of climate change before white messed everything up.
Not only is white bad, white is unhip. What do you want at your disco? White lights? No, black lights produce the right mood.
Let's fight the white and save the world!
That is ontopic. You try to read that page, then turn off the computer and leave internet for a few days. That is really power saving
Maybe PCstats should apply their own power-saving strategies to themselves (less CPU-intensive flash crap).
Anyway it appears only the CRT has a significant savings with White google versus Black blackle.com. LCDs gain almost nothing.
My AC stalker: " I personally agree with your posts most of the time, but that won't keep me from modding you troll"
The real power-saving web pages are simple and clean ones that that use the least CPU time to load, without bloated Web 2.0 javascript mashups of dozens of irrelevant sites and web bugs that keep track of you. TFA doesn't seem to mention that.
Firstly, I'm extremely skeptical of one of the conclusions - 'flash will make a CRT monitor use more power' - which I just don't believe - it will use an amount of power dependent on the average screen brightness - which may be an increase over black.
LCDs are different - the panel does actually take some energy to change state, and the lag compensation circuitry will use more in motion.
Secondly - a huge part has been missed out of this.
Power consumption of the computer.
Flash, or javascript, even in the background, can considerably increase power.
For example, I just closed all of the flash/animated things in the background on other tabs in firefox, and the CPU usage is now bouncing around 2%, with the computer using 17W.
If I start up a new tab with some flash, and gif animations, it goes up to 25W. (+8W)
Even switching away from the tab only takes it to 23W or so. (+5W)
It would be interesting to work out the total electricity wasted by common flash ads.
I don't really see the problem with "not insignificant".
Just because something is "not insignificant" doesn't make it "significant".
Say I give you a papercut. You'll be in a "not insignificant" amount of pain.. in fact, you'll probably curse me all day long.
But it's not exactly a "significant" amount of pain either.. it's not like you're curled up on the floor begging for somebody, anybody, to put you out of your misery or at least give you an OTC painkiller.
Perhaps a completely alternative term could have been used - suggestions?
( I used 'measurable' in another post - but while 0.01% might be measurable, it but would be insignificant. )
It would have been interesting to include the whole computer in the power measurement. How much more electricity is drawn by a javascript infested site than one that is just static HTML and images ? How much more is drawn if there are 100 components to build the page instead of 20 (don't forget to include the consumption of your broadband modem, etc, ...) ? How much more electricity does flash use ? How much more through heavy use of AJAX ?
The biggest difference that they showed was that the use of a glass monitor was about double that of a LCD. With an LCD the CPU/... consumption would be a bigger fraction of the whole thing.
Web pages with FLASH waste more power.
Democracy Now! - uncensored, anti-establishment news
That's not so strange in electronics.
Take FETs - undriven they're fine, saturated they're fine, but the Ohmic region you typically (when using it as a switch) want to stay out of because the FET's just going to burn the excess off in the form of heat.
There's a bunch of reasons why some regions may take more energy than others. I wouldn't know what the reason is for the panel they used, somebody more intimately familiar with driver design and panel response would have to chime in.
People still have CRTs?
How ridiculous.
Ability to display perfect black color;
Ability to display more than one resolution correctly (useful for games, old video card = new games at reduced resolution);
Ultrafast response time, no input lag;
Reliable and have long life (people saying things like "My LCD started acting weird, but it's 3 years old, time for a new one", while my 12 year old CRT works great), but can also be repaired if necessary (well, other than the failure of the tube obviously);
More affordable than a 24" LCD that can display 2304x1440 (if such a thing even exists);
Great image quality.
The only advantages of LCDs are size, weight and power consumption - all of these are not primary features of a monitor, at least for me (the same way that I don't buy a car based solely on the fuel consumption, or a computer based on its power consumption and size - I look for performance and cost first).
>> Voltage mutiplied by current in Amps equals Watts.
NO. For God's sake will people stop making this mistake.
Voltage multiplied by current in Amps equals VA, not Watts. If you want watts, you have to multiply Voltage in Volts, Current in Amps, and the cosine of the angle between them (which is more commonly known as the power factor.
VA = V*A
Watts = V*A*PF
But the characters are black, so each one represents energy saved.
Amps are how tall they are, volts how many are arriving in a frame of time or frequency.
Not that it affects the product (charge/energy), but amps measure transfer of charge over time, and volts measure electrical potential energy, so volts should be the height of the waves (gravitational potential energy) and amps the rate of arrival (in terms of volume of water per unit time, not waves per unit time).
It's a perfectly valid way to measure since we pay based on wattage per hour.
I don't know about where you're from, but around here we pay for energy in watt-hours (1 W*h = 3600 J), not watts per hour.
"The state is that great fiction by which everyone tries to live at the expense of everyone else." - Bastiat
The only advantages of LCDs are size, weight and power consumption - all of these are not primary features of a monitor, at least for me (the same way that I don't buy a car based solely on the fuel consumption, or a computer based on its power consumption and size - I look for performance and cost first).
The LCD advantage that I prefer? Not irradiating remaining eye.
I feel fantastic, and I'm still alive.
Set your background to black, and all your text so it only displays half the time.
I eat only the real part of complex carbohydrates.
CRTs have big geometry problems...
CRTs also take up so much space that is ridiculous.
Is that not also a geometry problem?
When our name is on the back of your car, we're behind you all the way!
Oh, so very, very much this.
Let me count the problems with light-text-on-dark-background:
1) If you have cataracts, corneal irritation, or smudged glasses, bright objects against a dark background are MUCH harder to resolve than dark images against a white background. With black-on-white, you just get reduced contrast; with white-on-black, you get distracting smears and rays all over the page.
2) In a dim room, your pupils dilate more if the scene before you is mostly dark, and dilated pupils generally produce poorer acuity. A bright background causes your pupils to contract, and just like stopping down a cheap camera lens, it improves the focus of the image hitting your retina.
3) In a bright room, a mostly-dark display will be more obscured by reflections and glare.
This is one reason I stopped hanging out at dpreview.com. Yeah, I know, photographers think their stuff looks better against a black background, but more than five or ten minutes on the site gives me a headache.
this is minutiae.....
3.8W is hardly a minute amount of power. If I did my math right, it's approximately the amount of power it takes to lift a full soda can (~390g) 1 meter in 1 second.
Let's say each Google query takes 10 seconds of viewing time, so you could save 38 watt-seconds per query by going black. Multiply this by 3 Billion queries per day, times 365 days/year. That's 12GWh (to 2 significant figures) of electricity that could be saved annually by changing a couple lines of code.
Power costs around $0.10/KWh. I don't consider $1.2M/year to be a minute amount of money.
Why is it that the proponents of "one nation under God" are so eager to get rid of "liberty and justice for all"?
>> Voltage mutiplied by current in Amps equals Watts.
NO. For God's sake will people stop making this mistake.
Voltage multiplied by current in Amps equals VA, not Watts. If you want watts, you have to multiply Voltage in Volts, Current in Amps, and the cosine of the angle between them (which is more commonly known as the power factor.
VA = V*A Watts = V*A*PF
No, Watts is really Voltage times Current. But when referring to AC systems, definitions get all screwed up. Just look at "kWh" - what a mess. It's like electricians have their own definitions for these units. I suppose it is understandable - using a single number to approximate a waveform and then performing calculations using Ohms Law makes most tasks much easier.
So pointing out the difference between Watts and VA is good - thanks for that. But don't be calling the real definition for Watts wrong. Also, your definition for power factor is not correct - or at least it is dated. It only applies to AC systems where the waveform is shifted. Power factor also applies to waveforms that are modified in other ways. For example, a computer power supply without power factor correction consumes pulses of power during the peak points of the sine wave. This changes the shape of the wave without resulting in a phase shift. With power factor correction, a control circuit draws power throughout the entire waveform so that the sine wave is not distorted.
I wonder what they used to measure power usage for this test. Did the instrument record true RMS power? Those instruments are much more expensive but required for accurate results. Guess I should rtfa.
No. The amount of fuel your power plant uses is proportional to the power it is supplying. In case of coal plant: there is less coal being burned, in case of hydroelectric: less water needs to go through the turbines, in case of nuclear: control rods are inserted into the reactor core to slow down the reaction of the fuel rods by absorbing neutrons.
Less power used = less power generated = less fuel used.
The above post has incorrect assumptions.
Standard LCD screens do not alter the intensity of the backlight based on the information displayed on the screen, and the backlight and it's inverter is the majority of the power consumption. In addition, the drive circuit that aligns the liquid crystals can work opposite from how you expect in a TFT. Most TN screens, for example, are white or light gray when unpowered - refreshing the pixels to a black state takes more transistor drive than making the screen white. This is the technology you will find in most portable devices and computer monitors.
Some LED-backlit TVs use dynamic backlight, or even zone-dynamic backlight, where (mainly to create ridiculous contrast ratio specs) the backlight is reduced to the maximum temporal white level needed, or for multi-area addressable systems, the brightest backlight needed in an area.
The only portable devices where the brightness of the screen data is directly related to energy consumption would be those with OLED screens (such as the Samsung Galaxy SII line). The individual pixels are miniature LEDs, and when a pixel is black, they are turned off. On these AMOLED display phones, a black wallpaper can use far less power.
When I think of "power-saving webpages", I may be more concerned about one that runs my CPU at 100% for several seconds to display a page, Slashdot.