PC Power Consumption

 

Introduction

With renewed interest in energy conservation, spurred by the ongoing concerns about the cost of energy production, as well as the impact of the heavy use of fossil fuels on global warming, it seems like now is a good time to look at how much power is consumed by today's computer hardware.  The notion that relatively small changes in our energy use can have a significant impact over the long run is exemplified in the current campaign to get people to adopt the use of compact florescent light bulbs, instead of incandescent bulbs.  Perhaps we can find opportunities for significant energy savings in how we use our computers, as well.

Previous survey research conducted in 2001, by the Department of Energy's Energy Information Administration, suggested that computer use did not represent a sizable portion of electrical power consumption in American households; the report buried PC usage in the catch-all category for energy usage, "All Others".  According to this survey, computers and their printers only accounted for about 2% of the average household's electricity use.

 

However, this survey is several years old, and it certainly doesn't represent the percentage of power consumption by computer equipment in my home.  While it needs to be granted that I make rather heavy use of my computer (or more accurately, computers, as I've even been known to have two or three running at the same time), other computer enthusiasts or "power users" may find this to be the case for them, too.  The number of households with computers has no doubt increased in the last six years, and I believe that it is no longer unusual for homes to have more than just one household computer.   

I've been of the school of thought that frequently turning electronic equipment on and off is relatively hard on it and tends to shorten its life span, more so than simply leaving it on all the time; so, my tendency has been to leave my computer and monitors running, except when I know that I won't be making any use of them for several hours.  Typically I start up my computer, along with its two monitors, early in the morning and leave it running all day until I go to bed.  I'll readily concede that this notion about the negative impact of turning electronic equipment on and off is more a matter of a prevailing opinion than a well documented fact.  It probably has been accepted more because it seems to make sense than because of any preponderance of documented evidence.  Certainly, I can't provide any quantitative data on how much of a deleterious effect we are talking about.  And, there are some counter arguments to be made, such as  how running your computer constantly tends to be hard on mechanical parts, especially the fans.  While fans are relatively inexpensive to replace, if they are running all the time, the inside of your case will more quickly accumulate dust, and dust can lead to problems both for the fans and the electronics.  So, without trying to settle this issue based on competing opinions, maybe it will helpful to simply look at how costly my current practices are.

 

Some Numbers

I used a Kill-a-Watt home power meter to see how much power my computer equipment was actually using.  This meter can display the current amount of energy being drawn in watts, and it can keep track of the total amount of power used over time, which is presented in kilowatt hours (kWh).  This is what the power meter outside of your house keeps track of and what your utility company uses to calculate your bill.  One kWh represents the power used by ten 100 watt light bulbs over the span one hour, for example.  

To begin with, I ran all my computer equipment through this meter, and over a period of several days, I collected the total kilowatt hours of energy consumed.  I found that my computer usage required about 7.34 kWh per day.  This is roughly the equivalent of leaving three 100-watt light bulbs burning day and night.  Multiplying this amount by 31 days results in a total of 227 kWh per month, which represents a significant portion of my home's total electricity consumption.  

There may be a tendency to view the cost of electricity as inexpensive, since we only pay a few cents per kWh.  For example, I pay 7.314 cents per kWh.  Returning to my previous example, this means that I can turn on ten 100 watt electric lights and leave them burning for an hour, and it will only cost me about 7 cents.  The thing to keep in mind, especially for appliances that are on most of the time, is that there are 730 hours in a month and 8760 hours in a year.  Doing the math, an appliance that averages of 1 kWh per hour will end up costing me $53.40 per month or $640.70 per year.  In the case of my computer hardware, which is averaging about .306 kWh per hour, this means that I'm spending around $16 per month to power my computer equipment or about $196 per year.

Below, I've looked more specifically at the power consumption of my various computer components, starting with the computer's main "box".  Although the first system listed, the Core 2 Duo system, is the one that I use all the time, I've got a couple of other computers that I'll occasionally use for various purposes.  I've included the power usage numbers for them in order to provide a small sample of how much power different computer systems might require.

Systems (Main Box) Watts
Core 2 Duo System with 2GB RAM and 6800GT video card
Idle at Windows Desktop 100w
Loaded (running Prime95 and Aquamark3) 176w
Gaming (Battlefield 2) 150w
Windows Stand-by Mode  81w
Powered Off and Hibernating   2w
Athlon XP System with 1.5GB of RAM and 9800Pro video card
Idle at Windows Desktop 148w
Loaded (running Prime95 and Aquamark3) 190w
Windows Stand-by Mode  88w
Powered Off   5w
Celeron 1200 (Tualatin) System with 512MB of RAM and Ti4200 video card
Idle at Windows Desktop  90w
Loaded (running Prime95 and Aquamark3) 105w
Powered Off   1w
Toshiba Portege R100 Laptop with 1.1GHz Pentium M, 512mb RAM, and 12.1" LCD
Idle at Windows Desktop 14w
3D Pipes Screensaver 18w
Virus Scan 22w
Stand-by 0w
Hibernate 0w
Powered Off, but Battery Recharging  4w

I was a bit surprised to see that my newer and more powerful Core 2 Duo system consumed less power than my older Athlon XP system.  I'm guessing this is due to a combination of factors, including a more efficient power supply and the fact that newer processing chips are smaller and require lower voltages to power them.  In any case, we can see that a computer typically will consume around 100 to 150 watts and that this amount will increase during periods of intensive usage.  

I also checked out the power usage while the computer was in the Windows controlled stand-by mode.  Interestingly, my older computer gained more savings from this than did my newer system.  Perhaps this was because they both dropped down to about the same 80w level in stand-by mode, but my newer computer's idle power usage was lower; thus, the newer computer's net power savings were less.

Although the amount of energy required is not very much, you can see how all these computer systems continue to draw a few watts of power when they are powered off.  This "stand-by" power is what makes it possible to start a computer up by means other than hitting the power switch.

The computer's monitor is another significant user of electrical power, especially in my case, since I'm still using cathode ray tube (CRT) monitors.  The fact that I'm still using these monitors is a good illustration of my belief that the extra money spent to cover the cost for a better monitor is worthwhile, since a good monitor will often out-live other computer components, which too soon seem to become dated and even obsolete.  Although I would have a hard time thinking up a scenario where I would recommend buying a new CRT monitor to anyone, my CRT monitors continue to work fine, and they even have a few advantages over LCDs, such as wider viewing angles, the capacity to display a range of resolutions well, and faster response times (i.e., the speed with which they can accurately change the image they display).  Looking at the numbers below, we readily can see that power savings is not one of the CRT's strengths.  A 19" CRT monitor uses around 100 watts, and the smaller 17" displays only somewhat less.  Although I don't have a LCD monitor handy to plug into the Kill A Watt meter, a quick check of the specifications for 19" LCDs indicates that they tend to consume something more like 35 watts of power when in use.  This is about a third of what a CRT monitor, which is capable of displaying a comparable resolution, requires.

Monitors Watts
Hitachi 751 19" CRT Monitor
On and receiving video signal 97w
Off but plugged in and Stand-by  2w
Sony E200 17" CRT Monitor
On and receiving video signal 76w
Stand-by  1w
Off but plugged in  0w
KDS 17" CRT Monitor
On and receiving video signal 56w
Off but plugged in and Stand-by  2w

I should mention that when the computer goes into stand-by mode, it no longer sends a video signal, and this means the monitor will drop into its stand-by mode, as well.  In this stand-by state, the monitor usually only requires a couple of watts of power, which often is equal to what the monitor continues to draw when it is turned off but still plugged into a power outlet.

Other peripherals, such as speakers, modems, and routers are not significant consumers of electrical power, as they usually only require 5 to 10 watts each.  Printers also only consume a few watts of power while they are in stand-by mode.  

Peripherals Watts
Yahmaha 2.0 Speaker System
Turned On   4w
Turned Off   4w
Microsoft 2.1 Speaker System
Turned On (no off option, other than to unplug)  10w
Alcatel DSL Modem   6w
Linksys Router (wired)   7w
OkiData OL400e LED Printer (similar to a laser printer)
When printer is warming up from being off or a standby state 400w
When idle (fan still running)  12w
Stand-by (fan off)  10w
While Printing varies from 30 to 400w

Although my printer doesn't consume much power in stand-by mode, it does use quite a bit of power during the couple of minutes it requires to warm-up and while it actually is printing.  The OL400e is similar to a laser printer; it fuses toner to paper during the printing process.  This is done using heat, which means it can draw a fair amount of power while printing.  Although I didn't measure the power use of an inkjet printer, I imagine that it might require considerably less power when printing.

 

Conclusions and Recommendations

As I looked at the above numbers, I realized that the amount of power that my computer use was requiring was a significant amount of my monthly utility bill, and therefore, the cost was significant, as well.  I also could see that when running my main computer, along with its two monitors, those CRTs were using more power than the computer itself required.  So, I've begun to rethink the idea of simply leaving my system running all the time.  I'm currently experimenting with having my computer go into standby mode after two hours of inactivity.  While the my computer only draws about 20 watts less power in stand-by, it does put the monitors to sleep, which saves me the use of 170 watts right there.

The problem with the automatic power saving schemes built into Windows is that they can only track periods of inactivity and use this as a basis for deciding whether to power down or not.  Ideally, what really needs to be known is how long a period of inactivity is likely to last, and only the user is in a position to make some predictions about that.  This means that it is pretty much up to me to make sure that my computer is shut down during significant periods of inactivity.  

I've decided to take advantage of Window's hibernation state.  When I think that I'm unlikely to be using my computer for more a period of a couple of hours or more, I'll manually power it down into the hibernation state.  This essentially is the same as turning the computer off, but it has the advantage of the computer being able to boot up more quickly afterwards.  Data currently in the system's memory is saved to the hard disk, when the computer goes into hibernation, and then it is restored to system memory, when the computer is restarted.

I suppose the equivalent to the current campaign to get people to change over from incandescent bulbs to compact florescent bulbs would be to change out our CRT monitors for LCD monitors; unfortunately, the cost of doing so is quite a bit more than it is for changing light bulbs.  So, for now, I'll see if being more conscious about my computer using habits leads to me being more conscientious about conservation, as well.  

 

Sequoyah Computer

2-26-2007

(Updated 4-18-2007: added Toshiba Laptop data.)