Netbook Battery Tests: Windows XP vs Windows 7

41 Comments

windows7-battery-testPrior to the final release of Microsoft Windows 7, there some question as to what the new operating system would do for battery life on laptops. At that time, I felt it was too early to make any judgments, simply because final hardware drivers and power management utilities weren’t available. Now that Windows 7 is available, most vendors have provided software updates, so it’s a good time to check in on this. I spent yesterday running a Battery Eater test on my Toshiba NB205 netbook and then re-ran the test today. I have the device setup for dual booting between Windows XP and Windows 7, so it’s pretty easy to run the test. Battery Eater stresses out the device under test to provide the minimum runtime you can expect, not the maximum. I tried to setup the environments as equal as possible in what I consider a reasonable “real world” situation: screen brightness at 40%, Wi-Fi on and all power management features disabled. I’m a little surprised at the results.

windows-xp-battery-testYesterday’s test with Windows XP yielded a minimum runtime of 5 hours and 5o minutes. I get around 8 hours of usage with proper power management and basic web usage, so the Battery Eater test seems about right. Today’s test with Windows 7 was about 10% shorter, however — the netbook battery died off at 5 hours and 15 minutes.

I’m scratching my head a little over this because I had read about a similar test using the ASUS Eee PC 1008. In that test, Windows XP ran for 5 hours and 3 minutes on a single charge. The same device with Windows 7 ran for more than an hour longer — 6 hours and 7 minutes. Since the Eee PC 1008 and the Toshiba NB205 are essentially built from the same hardware, the only conclusion I’ve drawn is from the one variable: device manufacturer software.

When I got the latest Toshiba Power Management utility for Windows 7 last week, I was leery that it wasn’t quite the same as it was for Windows XP. For starters, there isn’t a Toshiba application or GUI for the power management — it’s simply integrated within the Windows 7 Power Management applet in the Control Panel. And even there, I only see one bit of evidence that there’s any Toshiba optimizations.

The other possibility here is that the ASUS Eee PC test was flawed. I’m not saying it was, but it was performed with two different netbooks — one had a brand new battery while the other had been through a half-dozen charge cycles. Even with the different physical hardware, I don’t think the test could be skewed by 20%, but it’s possible.

Just for kicks, I think I’m going to re-run the test one last time once the battery is fully charged again. I was running all of the Aero bits under Windows 7, so I’m wondering how much battery life I can gain by turning those features off. If I find that the battery life gain is 10% or more, I’ll report back in, because I think that’s significant.

41 Comments

Kevin Shih

I own an eeepc 1000h
Windows XP lasts about 6h25m if i have the brightness down to min, wireless and bluetooth off, and no programs running in the background.
normal usage lasts around 4 hours. normal usage as in surfing the web, facebook and gmail. And occasionally youtube.

With Windows 7, i got probably at most 4h30m.
3h30m if i use it “normally”

But with Ubuntu 9.10, with proper power management, i can get around 8h55m

I also have a touchscreen mod, and it draws quite a lot of power. Maybe that’s why my computer doesn’t last as long as i want it to.

armandete

Ubuntu what? Please tell us how you did it, in ubuntu there are no “real” or optimized drivers, and generally it actually sounds more like an another linux-lover lie and/or idealism… I really, but REALLY don’t believe you. It’s great to like this or that OS, and defending it for it’s advantages, but linux people is just less and less credible with this kind of childness. That OR you’re having real problems in your Windows installations, because I’m getting the double in XP, about 7hs30min with light wifi use (intermitent). Almost 7hs in W7. And (like MANY pleople) about 3h30min to 4hs15min in Ubuntu. I even saw a blog saying “let’s admit it, Ubuntu’s battery life sucks”… so… stop defending your hippie society with blindness and admit that there are not ONLY advantages.

Andreas

BIOS version might interfer with battery life as well (with regards to the eee test where two different netbooks were used).

Jack

Did you know that BAPCo’s battery life measurement benchmark, MobileMark 2007 now supports 64 and 32-bit versions of Windows 7. You just need to download the latest patch 1.06 and your ready to roll. If you sign up on the BAPCo Press Network (BPN), you will be sent a press copy.

Armando

Hello one more time (i’m talking alone at it seems), just updated my BIOS to a new versions, asus support page said: Supports windows 7. It was true! Now with official drivers and bios i’m getting about 7 hours of use (same as XP) in my eee 1000 ha :) it was matter of time until asus supports windows 7, excelent! Sorry for my english…

armandete

Hello, me again, Just posting the good news (for asus (damn it… again.. Asus? Asus’s? Asu’s? English is not that simple ahh..) Asus just released ALL the official drivers for most EEE netbooks :) super good news right!? Maybe someone else can try battery life now? I think i’ll just do the same…

Great blog by the way, i’m happy i’ve found it

armandete

Hello! Sorry about my english, I just made some tests with my Asus EEE 1000HA that Microsoft gave to me lately… (Happy!), first, I was even scared about how Windows 7 eated my battery… I could check the % satus and see 80% then have a drink and it said 78%, horrible. It used to seems that the power saving options wasn’t working very well, specially because the Atom was in 1.6Ghz all the time… then i’ve installed Super Hybrid Engine (Asus’s (or Asu’s??) power saver software) and set it to “power saving mode”, and it ran with a bottom of 650Mhz, but going to 800Mhz and very eventually 1.2Ghz… never needed to go faster as it worked very smootly… battery time did change a lot, BUT even then I still have a significative diference where XP runs longer, about 20 – 50 mins depending on what I’m doing. 20 – 50 mins means the time to get to the 50% battery level… I don’t use to pass that point anyways as it still a long duration battery…

Well I hope you forgive my english, if is too bad next time I promisse to use google translator :)!

PD: I’m going to try MOBLIN right now, it’s downloadint at this time… if I see any interesting advantage in matters of battery life, i’ll let you know ;)!

Kevin C. Tofel

“Testing battery life by disabling the operating systems power management features is pointless.”

I like to know the barebones *minimum* amount of runtime I could possibly get from a device when I leave the house. This test does exactly that. Pointless to you, but not to me.

Your description for a relevant test is a totally different type of test (not any less valuable, by any means, but different) — one that would provide results for the maximum runtime.

kubel

This test is like comparing an 1989 Mustang to a 2010 Camero by removing their engines and pushing them down a hill. How fast they go is completely irrelevant when what matters is disabled.

Testing battery life by disabling the operating systems power management features is pointless. Also, third-party power management utilities can interfere.

For a relevant test, both systems must be identical (aside from the operating system), both systems must have third-party utilities removed, and both systems must be set to their most aggressive power saving features (without entering hibernation or sleep mode).

Battery Eater is designed to test the battery itself, not the operating system power savings.

bj

Just to add more fuel… have you considered adding a 3rd OS to the mix? Especially since the main contenders in the netbook space are going to be XP, 7, & some form of Ubuntu.

I have noticed that with the recent upgrade to Ubuntu 9.10 I am getting longer battery life out of my 1410. With the crappy NA battery it is still too low but 9.10 seems better than 9.04.

Cindy

Would you recommend at this point someone buying one of the available inexpensive XP netbooks or waiting for a 7 netbook?

webtax

i too have dual boot xp and seven, in a msi wind netbook.

been using xp for a year, now used seven for a week, and now back to xp.

i can say for sure that , at least with the software i used, i get a significant more battery time in xp. Also the performance is better.

webtax

by the way, i disabled aero, composition, and visual styles.. so it looked pretty much like win95 :P.

David

Thanks for the info.

Yes, I really think the added UI graphic features with Aero results in greater power consumption and to me turning it off means going backwards to XP.

One question I hope someone can point me right direction is “How about running 64bit. Would it means greater performance, efficiency, less power consumption?”.

Kevin C. Tofel

Good question. I’d suspect it might offer a little of what you’re suggesting, but I can’t test it on the netbooks I own. The current Intel Atom N270 / N280 isn’t 64-bit capable. The upcoming N450 / N470 that we should see in a few months, however, is.

Jake

I’d be interested to know whether Aero is responsible for the difference. Assuming that you have proper WDDM 1.1 drivers then I would expect that the performance and memory management gains would reduce power consumption.

EdH

I agree with Scott. I don’t understand the point of this testing methodology. if you want to test raw battery life between multiple machines, then yes, disable all power management features and see how fast they run down.

But to test the power management capabilites of different operating systems by disabling those very features? Doesn’t make sense to me.

I am not sure exactly how to go about doing such a test, but disabling the very features that enhance what you are testing doesn’t make sense.

Would be like testing some hybrid vehicles for best mileage and then disabling all electrical power systems so you are only running on the gas engine to get a true reading of miles per gallon.

Scott

Kevin

I’m confused. You start out saying that “all power management features disabled” – and then site the Toshiba power management software for the difference???

Also wouldn’t the (better) power management features be how Win 7 could beat XP? It seems odd to test for max battery life without power management.

Kevin C. Tofel

Scott, you should be confused by what I said — because you’re absolutely right! This is what I get for multitasking during a World Series game! ;) I was looking so hard for a variable that I didn’t think to see if it made sense. Your point leads me to believe that HG is right. Aero is surely hurting battery life, which we already knew from Vista. I’m suspecting more and more (as Brad does) that the 1008HA battery test results between XP and Win 7 are incorrect. Thanks for your point and I promise not to write during any future World Series games. :)

James Kendrick

Totally OT but it was a great game in Kevin’s defense. Lee’s pitching was a joy to watch. I felt so badly when a throwing error cost him a well deserved shutout. OK, back to tech, now.

David T

Battery run time test is actually not a consistent science (Take note from the enthusiasts who are obsessed with testing run time on different batteries on different flashlights (to an expert like level), and you can see there are variance in just about everything. In their respective field, they plot discharge curve of the battery / cell over time, as well as measure the brightness output of the flashlight as well. A flashlight (even the newer electronically controlled LED ones) is much simpler than a computer, and still has lots of variations. One can only do a lot of repeated tests and compare results in order to minimize variances (statistical analysis).

But if one must ask why, then I’d speculate that Windows 7 has a higher CPU resource requirement that makes the CPU runs more. In today’s CPU, they are very dynamic in terms of power usage vs. load, so that it can preserve power when the load is light. It is only natural that Windows 7 has a higher load.

Another thing to look at is HDD activity. Does one OS use the HDD more often than the other? On this one I’m not very sure since Windows 7 has fairly intelligent pre-fetch compared to XP, as long as there is enough RAM. Was the swap file used often or not? etc.

I think power management can only do so much, if there is active usage of the computer. So I doubt (but I’m not sure) if the power management utility is to blamed at this early stage. I’d put my bet on the OS itself first. Windows 7 just simply has more stuff.

In my Acer Aspire One 9″ with CF mod, I didn’t pay much attention to this, but it seems that the run time is probably about the same on both OS (XP and Win 7), but as I haven’t really paid attention, I don’t really know if one can be 25% better or not. They both get less than 2 hours on the 3 cell pack (which is not enough for me), so there is no need to compare for my own sake.

HG

I have a feeling that since Xp doesn’t run with Aero and Win 7 does and the test was done with Aero on. That might be the reason for lower battery life on the NB205. I’m also going to try and turn off Aero and just see what kind of battery life I get. Let us know what kind of battery power you get with Windows 7 and Aero off in your test. I have a feeling that when you turned your back your cat pulled your battery out. ;)

tsog

how about doing a real world test (surfing) for xp and 7 on the same computer and compare the difference in battery lives?

also, i think in xp, u can watch ur power consumption in real time by using the performance app (go to control panel -> classic view -> administrative tools -> performance). here’s the help topic:

Monitoring Battery Status
The Battery Status performance object monitors activity on laptop batteries. The counters included in this object are:

ChargeRate. The rate of charge in milliwatts (mW). If the battery is not charging, the value of this counter is 0.
DischargeRate. The rate of discharge in mW. If the battery is not discharging, the value of this counter is 0.
RemainingCapacity. The amount of charge left in the battery in mW.
Tag. The number of times the battery was changed.
Voltage. The voltage of the battery in millivolts (mV).

if u can find out the battery consumption for the same netbook under xp and 7, perhaps u can find out why win 7 uses more power

tsog

finally found out how:
1. in performance, click the “+” icon (should display “Add” when u hover over it
2. make sure u r on battery (not plug in)
3. scroll down the list under performance object and select battery status
4. select variables u want to monitor from the list of counters (i.e., charge rate, discharge rate, etc.)

u should get a graph of ur current rate of consumption, etc.
for reference the discharge rate for my p1620 at max battery and 3/8 brightness, idle, is about 10000mw, or 10w

matt

with my MSI wind, i’ve noticed that a “full charge” can vary a lot. i use BatteryBar, which has a few really nice features: showing the current Discharge Rate, the current capacity, and the Full Capacity. one day the full capacity 21,168 mWh while the next will be 21,790 mWh. since this variation occurs, a better test would be to compare the Discharge rates between the OS’s.

I have the RC for 7 and XP on my wind right now, with Battery Bar installed on both, and in limited tests i’ve noticed that Discharge Rate to be higher in 7 consistently, whether it be playing videos, music, using OneNote, whatever. for example, playing an 80 mb, 22 min long video in x.264 in XP with wifi off, minimum screen brightness, causes a discharge of approximately 13,000 mW. in the RC for 7, doing the same causes a discharge of approximately 14,000 – 14,500.

granted, i didn’t have any MSI power utilities installed under 7(do they even have them?), but i would suggest looking at the Discharge Rate between the two OS’s for a good comparison. maybe play a video, have wifi on, screen brightness at 40%, and see how much the battery has discharged with BatteryBar (you can do this just by looking at the current capacity at the beginning of the video, and the current capacity at the end of the video). do this with both XP and 7 and compare the differences. while this may not give as good an estimate for the minimum runtime Battery Eater does, it will at least give you an idea to compare the 2 (and you won’t have to wait for hours for the battery to discharge twice).

Kevin C. Tofel

Excellent suggestion, Matt. I used to install a similar program to the one you mentioned – Notebook Hardware Control. It too shows the real-time discharge rate and even graphs it for later viewing. I may have to go back and look at that approach. Thanks!

Brad Linder

I tried to repeat the test cited on my site and haven’t been able to. I ran Battery Eater on a 1008HA review unit a few months ago and got 4 hours of battery life. I ran it again this week with a Windows 7 review unit and got just 3 hours.

I’ve asked Chris to clarify what settings he was using on Battery Eater, but he hasn’t gotten back to me.

While I’ve pretty much stopped using Battery Eater for most of my reviews, I am using a Toshiba NB205 demo unit that shipped with Windows 7 Starter right now. I’ll run the test this week and let you know the results to see if they differ at all from yours.

As far as I can tell, Windows 7 offers more power management settings than Windows XP… but I haven’t seen much evidence that it actually improves battery life on netbooks. In fact, in my limited testing it seems like it might have the opposite effect.

Kevin C. Tofel

It’s difficult to say. I think the XP version of the Toshiba Power Management utility offers a greater level of customization while the Windows 7 version is just a minor tweak to the native power management. I’ll take some screencaps of the XP utility and shoot them to you so you can see what I mean. :)

Bill

Kevin,

Does the Win. 7 Toshiba Power Management Utility allow you to set the cooling for the processor like the XP version does?

Bill

Brad Linder

I ran Battery Eater Pro in Classic Mode on a Toshiba NB205-330 that shipped with Windows 7 Starter today. The end result? 5 hours, 16 minutes.

So I’d say your test was pretty accurate.

Brad Linder

Well, at least we know the batteries are consistently built! (And that the battery eater test is reliable… if not exactly reflective of real world use).

PreHome

ur confusing the 2 machines, 1005 is the equivalent to ur machine not the 1008.

I think his test is flawed anyways, the 1008 doesn’t get even close too that much battery life (I own 1)

Kevin C. Tofel

Both (all three actually, although the 1005HA has an N270 or N280 CPU option) have the 1.66GHz Intel Atom N280 and Intel GMA 950 and 5400 RPM hard drives. Aside from the battery capacity, what’s the significant difference that you think I’ve confused?

I’m not sure it really matters because I’m not comparing the actual run time between my Toshiba and his ASUS. I’m comparing the battery life on my netbook with two different operating systems. ;)

H1N1

besides the battery capacity? thats pretty much the ENTIRE reason the times are different. the 1008 is a polymer based machine. the 1005 is the “sister” machine to the Toshiba, component wise. i realize your testing the OS difference not machine difference, but the reason he made the comment was because of your quote

“Since the Eee PC 1008 and the Toshiba NB205 are essentially built from the same hardware, the only conclusion I’ve drawn is from the one variable: device manufacturer software.”

besides, you didnt even acknowledge battery cap difference. you treated the machines as equals & only mentioned software difference as the variable.

Kevin C. Tofel

Um…for the purpose of this post (read the title), I’m comparing the difference in runtime on the same netbook with two different operating systems. The variable in my test is the OS. In this case, it doesn’t matter if I compare my netbook to a Dell Adamo because I’m not comparing runtimes between machines.

Let’s look a little deeper at the quote you pulled.

In his test (which is essentially what I tried to replicate), he saw a 20% increase in battery life with Win 7. I saw a 10% decrease. Neither he nor I changed any variables aside from the OS. So again, there is another variable at play in our tests when looking at the *percentage change in runtime for our respective machines*. I could care less if his battery has a different capacity because I’m not comparing my runtime to his. He’s comparing his two runtimes with different operating systems. I’m comparing my two runtimes with different operating systems. We are *not* comparing to each other, so it doesn’t matter.

Unless you’re saying Windows 7 runs longer on netbooks with lithium-polymer batteries than it does with XP. ;)

Comments are closed.