Mac and PC Power Consumption

I’m thinking of buying a new computer for home and thought I’d actually check power consumption (for a change) after noticing that (1) my work PC never goes to sleep despite my settings (I think it’s because it has a web server running on it as a “service”), and (2) a rather nice Gateway (Acer) tower I was looking at had a 750W power supply… I mean, 750W is enough to run a fan heater on low, or the same heat output as ten people.

Here’s what I’ve learned so that you don’t have to bother.

Computer Idle (Watts) Max (Watts)
Mac Pro (original) 4×2.66 (1) 171 250
Mac Pro (2008) 8×2.8 (1) 155 318
Mac Pro (2009) 8×2.3 (1) 146 309
Mac Mini (2009) 2×2.0, 9400M (1) 13 110
Mac Mini (Late 2006) 2×1.83, GMA950 (1) 23 110
24″ Intel iMac (2, 3) 80-100 135 (or 240?!)
PC Intel Core 2 Duo E6400, 8600 GTS (4) 87 163.5
PC AMD Athlon 64 X2 5000, 8600 GTS (4) 102 210
MacBook 2.4, 9400M (5, 7) 9W (8) 90W
MacBook Pro, 9400M+9600M (6, 7) 10W (8) 100W
Here's a graph in case you find digesting tables of numbers as unpleasant as I do.
Here's a graph in case you find digesting tables of numbers as unpleasant as I do.

Addendum

My Kill-a-watt showed up (I got mine from thinkgeek.com). So far my very rough results are that my Macbook Pro (late 2007) uses about 30-35W on “idle” (but with the screen full brightness) and the most I’ve gotten it up to is 68W, but I suspect it could get to around 90W if I could max out the GPU and the CPUs simultaneously.

It will be interesting to measure the power usage of some flat screens to see how well a Mac Mini compares to the iMac (and, at the same time, how well the iMac compares to modular options).

Incidentally, the Macbook Pro’s power brick consumes 1W when it’s not attached to the notebook, and around 24W when it’s charging the notebook but the notebook is asleep. The figures above were for a notebook whose battery needed recharging — it’s probable that power usage goes down if the battery is fully charged.

Notes and Sources

  1. Apple Computer’s tech support pages
  2. From this thread; varying peak power consumption figures probably reflect different GPUs/tasks and bad methodology
  3. Power consumption includes built-in 24″ monitor (varying idle power figures based on brightness)
  4. The Truth About PC Power Consumption on Tom’s Hardware
  5. MacBooks have a 45 Watt-hour battery and have a quoted maximum battery life of 5h when using wireless, so we can safely assume idle power consumption will be around 9W.
  6. MacBook Pros have a 50 Watt-hour battery and a quoted maximum battery life of 5h, so 10W.
  7. Guesstimate based on assumption that peak power consumption will be about 10x idle — which seems like it’s roughly right based on the Mac Mini’s peak power consumption.
  8. Edit: rereading this post I realize there’s a major boneheaded mistake in it. Since notebooks never draw power directly they will always draw (significantly) more power at the wall than they actually consume, and batteries of course take more power to charge than they later provide. I’ll update my figures and graph when my “Kill-a-watt” arrives.

Conclusions

I just changed the energy saver settings on the Mac Pro I’m using. Good grief!

The average cost of residential power in the US is 11.03 cents per kWh (Department of Energy statistics), so that means a Mac Pro on idle costs about $0.39 cents per day, or $141.07 per year, versus $0.04 and $12.56 for a Mac Mini (and probably a similar value for an iMac which puts its display to sleep). And note that the “typical” PC consumes over half the power of a Mac Pro while offering dramatically lower peak performance (as the Tom’s Hardware article might put it — if you were working on a video project that required 8h of manual work and then a bunch of rendering, the 8h would be slightly reduced on the Mac Pro vs. the PC (which would be running close to idle) and then the Mac Pro would finish rendering in one-quarter the time and go to sleep.

Oh, and by the way, if you buy a cheap PC and keep it for two years leaving it switched on most of the time, you’ll have paid an extra $200 for electricity compared to a Mac Mini. As I said at the outset, my work development PC is running a web server as a “service” and thus will never go to sleep.

If you want to save power (and money), get a Mac Mini or a notebook (assuming Windows notebooks are similarly frugal) — unless you need serious horsepower, in which case Mac Pros start to look good (as long as you put them to sleep when you’re not using them). If you bought a Mac Pro because you think Apple is “green”, but you only use it for fairly minor stuff and you don’t put it to sleep when you’re not using it (heck, even if you do) then you’ve made a grave error — your Mac Pro is using as much power on idle as a typical PC under heavy load. Mac Pros are great in terms of computational power per watt, but only if you’re actually using them.

  • Wow! Thanks for that Tonio 🙂

  • David

    Great piece. One question: when looking at the mac mini and the pcs’s did you include the power draw from the monitor (which is separate in those, but is not separate in the laptops and mac pro)? If so, awesome. If not, the #s are a bit misleading.

  • I should include power consumption for various “typical” displays to make Apples to Apples comparisons easier. I’d prefer to measure power consumption sans display across the board since displays are interchangeable. The iMac’s power consumption is particularly impressive given that it includes a (very bright) display. Aside from the iMac and the Macbooks, all figures are for the computer without display.

    One thing to bear in mind about notebooks is that they’re rather inefficient when used on battery power since it appears to take about 3 minutes to charge them for 1 minute of use (so each watt they’re using on battery power was 3W at the wall).

  • The Wolff

    Hi! I read your post about power consumption – actually found because I was looking for the “real” values for a Mac Mini (’09).
    I have a kill-o-watt hooked up to my Mini for constant measurement – however, I could only achieve those 13 watts if it was sleeping, not idling. When idling 98% I read 26 watts. Twice as much. Have you verified you values or you used the information available on websites? Just want to find out if I’m the only one with “bad” result.

  • I’m relying on published figures — I don’t have a Mac Mini. Is its power consumption 13W when sleeping? That would be terrible.

    I’ve seen other similar complaints: http://forums.macosxhints.com/showthread.php?p=546153 — but most comments about the Mac Mini’s power consumption being in the 20-30W range relate to the old G4 model, not the Intel minis. Of all the computer vendors Apple has historically been the most conservative when making quantitative claims (e.g. they were one of the last companies to refer to 13″ monitors as 14″ — they measured usable area while everyone else measured the picture tube) — in large part because they are frequently the target of lawsuits.

    What do you mean by “idling 98%”? If you mean 98% CPU, then — by definition — that isn’t idling.

  • Macko

    And what about the Atom and Nvidia ION platform? Max power usage 55W. But its not designed for “hard work”. Just as a home pc.

  • The Atom and nVidia ION platforms are certainly going to be contenders for low power among commodity PCs. Toms Hardware’s extensive tests showed that Core2 Duo delivers a lot more computational power per watt than the the Atom. The nVidia ION platform consumes 20W on idle, which is not particularly impressive compared to the Mac Minis which use less power on idle and yet deliver serious horsepower. (It also shows just how well engineered Apple’s hardware is — Apple isn’t “just” a software company.)

    Indeed, the Mac Mini just looks better and better — and with a 2.5″ SSD drive its main shortcoming (slow drives) would be eliminated, and power consumption further reduced.