Web Surfing Test Shows SSDs Better for Battery Life

A few days ago, we published a story about how much better our Eee PC 1000H performed when we swapped the system’s 5,400rpm Seagate Momentus hard drive out for a Samsung SATA II SSD drive. Not only did the system boot faster and all apps load faster, but we got 20 more minutes of battery life with the SSD. However, when we posted our story, we heard from users that Tom’s Hardware, a site we admire a great deal, recently published an article which claims that SSDs use more power than traditional hard drives. The Tom’s Hardware story got some major attention from sites like Engadget and even got a responses from SSD-makers Super Talent and Micron, who both claim that the drives used in the Tom’s test are “early generation” and therefore more power hungry than newer models. Is Tom’s Hardware right? We don’t think so. Some Engadget commenters found some serious flaws in Tom’s testing. A user named Basic made a point we totally agree with:

What we need is a battery-loading benchmark which is representative of REAL-WORLD USE. For example: boot up the machine with maximum charge at a controlled temperature and sit it on 5-6 auto-refreshing controlled content web pages. Reduce all the other power outlays to the minimum – Vista power saver mode, turn down the brightness, but maybe keep the wireless network connection just to keep it close to a real use-case.

Toms didn’t do that – they ran a benchmark designed to “keep the notebook busy” and surprise, surprise the SSD’s died faster under load – disproportionately to the power they consumed (perhaps NOT disproportionately to the work achieved in that time… but Tom’s doesn’t tell us how much “work” was achieved relative to the drive’s performance… they don’t even mention the power-saving settings.

At LAPTOP, we actually just developed a battery life test very similar to the one Basic mentions. Our Web surfing test is a simple Windows shell script that cycles through a series of 60 popular Web sites, loading each page and then pausing for 30 seconds to simulate a user reading the page. The browser cache in Firefox is disabled so that the entire page must be downloaded each time. Meanwhile, the script writes to a timelog file every 5 seconds. The test is conducted at the laptop’s default power saving settings, except that nothing — not the hard drive, the screen, or the CPU is set to go to sleep, and all battery alerts are disabled. Where we ran our Web surfing test on an Eee PC 1000H before, we decided we’d try it on a regular notebook. So we took a Gateway T-6828 with Vista Home Premimum SP1 and ran our test with its default Western Digital Scorpio WD2500BEVS 5,400rpm drive, a Samsung SATA II 64GB SSD, and a SanDisk SATA 5000 32GB SSD. Our results are below.

As you can see, both SSDs lasted 3:23, exactly 10 minutes longer than the default hard drive. While 10 minutes out three hours is not a significant gain, it’s certainly not a loss. In other words, the SSDs we tested don’t use more battery life than traditional hard drives, at least not during Web surfing.

To be fair to Tom’s Hardware, most of the drives they tested were different than ours. They used a Crucial, an Mtron, a MemoRight, and a SanDisk SATA 5000. We also used the SanDisk SATA 5000 as one of our test drives, but that’s the only one our tests have in common.

It’s noteworthy that, in the Tom’s Hardware Mobile Mark test, the SanDisk SATA 5000 drive got only one minute less battery life than the traditional hard drive, while others like the Crucial got as much as an hour less. As Basic noted, it looks like the higher performance drives were penalized by Mobile Mark because they managed to do more work. The SanDisk drive actually has poor transfer rates in compared to the other drives in Tom’s test so no wonder it got better battery life. Because our test is not performance intensive and involves an equal amount of work for all drives, both SSDs lasted the same amount of time.

Our conclusion is that, in real-world use, SSDs offer a small improvement in battery life. While this tiny improvement may not be enough to sell users on SSDs as power-saving devices, it is certainly enough to say that upgrading to SSD will not cost you any battery life and may provide you with more productive minutes as you wait shorter periods of time for programs to load or for your system to boot.

Update: Bar graph reformatted to start at an X axis of 0:00. Previous version of graph showed same exact numbers, but was a “zoom in” that showed from 3:07 to 3:28. After reading some user comments that this was confusing, we decided to change the axis lines to start at 0:00.

Tags: ssd, Storage
AUTHOR BIO
Avram Piltch
Avram Piltch
The official Geeks Geek, as his weekly column is titled, Avram Piltch has guided the editorial and production of Laptopmag.com since 2007. With his technical knowledge and passion for testing, Avram programmed several of LAPTOP's real-world benchmarks, including the LAPTOP Battery Test. He holds a master’s degree in English from NYU.
Avram Piltch on
Facebook Twitter Google+
LEAVE A REPLY
Name*
Email* (will not be published)
Website
*Indicates required field
Comments*
Submit Comments

  1. jjj Says:

    one thing that must be mentioned is that not all HDD’s are equal and there are plenty of drives that use less power then the WD Scorpio 250G
    so your conclusion that “SSDs offer a small improvement in battery life” it’s not verified yet
    maybe you could repeat the test with more hdd’s?

  2. Meh Says:

    While I agree that the previous testing was flawed (summarized well by Basic), I feel that this testing too might not be that good as a test of SSD battery consumption. While, yes, this might be a representative test of what a netbook or a laptop might be used for in everyday life, it is not a test which is entirely relevant to the unit you are testing.

    What you are testing is a solid state drive, a harddrive. The problem with the testing you have conducted is that it does not test how a SSD consumes power under load, but rather when not under load. If you have a script which checks a couple of sites at timed intervals, you have a great check on how the system as a whole consumes battery, but not one which is relevant for the harddrive. Since the website isn’t put in the machine’s cache, you are barely using the harddrive at all, meaning that the test as a whole will be flawed. If you only load you have are the websites in question, you will in all likelyhood use mostly ram. If we wish to test long-term performance under use, rather try to put a playlist of songs up, turn off speakers, and allow it to run untill the battery is gone. This would ensure that the drives are being used, as is relevant to the test.

  3. JoeBLogg Says:

    The graphs are so misleading… the x-axis (time) doesn’t start at zero.
    Seriously guys… stop manipulating visuals to mislead

  4. steve seguin Says:

    I think the only way I could prove this to myself, one way or another, is to get my hands on a set of relevant SSD drives.

    Seemingly though, what becomes clear is SSD efficiency depends on the application and type of SSD. Windows Vista may constantly keep the drive in use, if for example, it is indexing the drive constantly. If the page file is also heavily relied on, the SSD drive might not be able to get into an idle mode either. Also, SSD drives do not have a cache memory from what I know about them, so a hard drive might benefit from this under certain scenarios. As well, sequential reading from a harddrive, such as watching a movie off the harddrive, might reduce the hard drive’s power consumption when compared to random reads– which SSDs are better with.

    Curious, why not just buy more RAM, reduce the need for a paging file and use the RAM to store cached internet files for that session? Battery life for websurfing would then become a matter of which consumes more at idle.

    Anyways, until SSDs come down in price, their relatively equal performance to hard drives keeps me away. (mind you, an 8GB SSD is attractive in its own way)

  5. So, who's paying your bills? Says:

    Interesting choice on the graphics.

  6. dognip Says:

    P.S.

    There is nothing misleading with the graph!!!

    The drives are all equal to the 3:07 mark.
    It’s just like zooming in on it to show up close the end result difference!

    Just like a photo finish of a horse race.
    They don’t show the whole race!
    Just the nose at the finish line…

    The difference is not HUGE, but at least it’s not worse.
    The vertical lines are in increments of 7 seconds.
    A win by a mere 10 seconds.

    The point is that the SSDs are slightly better in that one factor but overall, they are just a better product for a Main drive for my system.

  7. dognip Says:

    Sorry, minutes…

  8. think Says:

    Meh, I don;t think the test if flawed. I think the keys to any test are to 1) vary only one thing (the HD), and keep everything else the same, so you can guage the effect of the HD differences, and 2)use a scenario or scenarios that represent real-world uses(s). As others point out, it is unclear at best if Tom’s Hardware only varied one thing in their test, and it seems like they changed the workload and the HD, so their test does not seem to isolate the impact of the SSD.

    In LAPTOP’s case, I believe that LAPTOP varied only one thing, and used a real-world scenario, so the test seems valid. Meh, you are also right that there are other scenarios worth testing that will likely reveal something, but that does not make this test invalid.

  9. Alien8 Says:

    Ten minutes of time difference and almost the triple size in the diagrams.

    Very good attempt:)

  10. TUX Says:

    I think the best way to check the power consuming by HD is to run anti virus scan this way you will insure that th Disk we be running all the time

  11. JoeBLogg Says:

    lol graphs have been updated. …

  12. BKM Says:

    These people Know what they are doing. If there was any relevant gains with this product we would be the first to Know. This is sad..
    The MFGs need to supply real test results. ( kick your PR people in the butt )

  13. Meh Says:

    I read through my own post, and I can see nowhere that I say that the test is invalid. It is a good test for the system as a whole, but the tasks done throughout the testing does not use the harddrive to a significant degree, which should be a prerequisite for a test of a harddrive.

  14. butter Says:

    @dognip: Well, you could also show the capacities of the drives on another bar graph – 32:64:250, then ZOOOOOOM in on those. If the time bars can show a triple size bar for a ~5-10% difference, then the capacity bars should look even more ridiculous than this graph.

    I noticed that they fixed the charts though since – good on them.

  15. Prakash Says:

    I agree that this test is flawed as the solid state disk is hardly used when you refresh a web page particularly with laptop with lots of ram. This test needs to be improved to take into productivity(time) * Power consumption(in watts). A typical usage of laptop would be internet surfing, listening to music watching video, making ppt, doing some work in word and excel, copying files. But this is very tough to simulate in a bench mark!

  16. Roco Says:

    Bottom line on power so far: don’t expect order-of-magnitude benefits from SSD. Everything “depends” on the app being run. Hyping the power savings of an SSD is simply a marketing way of doing things. Systems are complex (we know that) and SSDs are not revolutionary in any way (they have existed for a while — 20 years ago as your ROM). Small gains are to be expected in an evolutionary way. I wonder if the performance claims should be tested as well. My USB keychain runs at 1MB/s for sequential reads. My cheap disk runs at 50MB/s!!!

  17. Avram Piltch Says:

    One thing I think I neglected to mention about our test is that it closes and re-opens Firefox for every Web page it visits. It also writes several log files to the hard drive:

    1. timelog.txt – an update timestamp of how long the test has been going
    2. timelog2.txt – a copy of timelog.txt that is written 5 seconds later. This is to make sure that one of the timelogs survives in the event that the battery fails at the very moment timelog is being written.
    3. A new log file for the web site being visited. This means that a new .txt file is written every time firefox opens.

    So if the browser visits 400 web pages and lasts 3:23, the drive will have:

    written 400 web surfing log files
    opened firefox 400 times
    overwritten timelog.txt 2,436 times (once every 5 seconds)
    overwritten timelog2.txt 2,436 times

    I don’t know if that’s enough activity to satisfy everyone, but it is a constant level of activity.

  18. sunny beach Says:

    Buried for a myriad of misleading elements, including a heavily misleading graph (Excuseme, LAPTOPmag, but graphs always start at 0×0).

    I wouldn’t be surprised if LAPTOPmag was paid off to counter the Tom’s Hardware test, which was also flawed.

  19. Toms is a joke. Says:

    How can you admire Tom’s? I’ve hated their biased-often wrong nonsense ever since the wonderful days of their “confidental” 3.6ghz Pentium 4s.

  20. lee Says:

    remove my last one and just post this (there is plenty on disk actively there)

    the performace you get from SSD neglects any more or less batt is used at the time, Vista needs SSD due ot it been hard disk intensive,(on SSD you should turn off the defrag as it may shorten the life of the SSD, SSD does not need defraging due to all parts of the disk been at the same speed)

  21. Bandito Says:

    It seems to me that you may have accidently done the same thing that Tom’s Hardware did by not reporting the number of web page visits and Firefox loads.

    If the SSDs are indeed faster than the mechanical HD, they should’ve been able to complete more visit/load cycles in the same amount of time. Since you have logs of the number of cycles completed, they should also be factored into the equation, which would likely show an even higher performance rating for the SSDs.

  22. Avram Piltch Says:

    Bandito said:

    >It seems to me that you may have accidently done the same thing that Tom’s Hardware did by not reporting the number of web page visits and Firefox loads.

    Actually, the speed of the hard drive would have no effect on our Web surfing test, because of two things:

    1. The system pauses 30 seconds after loading each page to simulate a user reading the page and then pauses another 3 seconds after sending the “taskkill /IM firefox” command to ensure that firefox has closed before attempting to load the next page.

    2. Having a faster hard may cause Firefox to open a tiny bit faster, but it’s not going to cause the web pages themselves to download any faster from the Internet.

    It goes without saying, though, that if the battery life is longer then it is going to end up visiting additional web pages in the extra 10 minutes of life. I’m happy to go back and read the logs and tell you how many web pages each hard drive visited if that helps.

  23. netbsd Says:

    It would be interesting to repeat the experiment using SDHC cards. You may recall that the Psion netBook only had a CF and PCMCIA slot, not a disk, solid state or otherwise.

    Maybe a fast SDHC card is better than a slow SSD and many of us might as well buy cheaper (2 SDHC slots, no SSD) netbooks.

  24. Bandito Says:

    It would be interesting to see the numbers just for curiosity’s sake. The HD and SanDisk should be pretty close if the run time difference is factored in, but I’d expect a bit of difference on the Samsung.

    I guess that there are still many factors to consider, such as how much of the program is actually being read from the drive as opposed to being drawn from cache memory, the possibility that cache memory erases any real performance differences between the drives when writing the small log files, and the fact that the load time of the web pages will vary due to traffic load and network routing variations. It seems that something that appears to be simple starts to get complicated very quickly!

    Still, it’d be nice to know what you got.

    Thanks for providing a rebuttal to the Tom’s Hardware article. It didn’t seem to quite add up correctly to me either, with what I know of electronics and physics. The other thing they failed to mentioned was how they did the power measurements that they published. Granted, the specs on the fastest SSD drive showed it to be rather a power hog, but their conclusions were a bit presuming and too broadly generalized to all SSDs.

    Thanks again!

  25. amadlopes Says:

    Both tests are ok. This tests a system with low charge and the other from Tom’s Hrdware tests a system with high charge. Choose form yourself. If you only use the system to webfurfing and other things that don’t too much the disks than buy it with ssd, if you use apps that use much the disk well than buy one iwth normal hard disk. This is like asking what computer do i have to buy ? We have to know first for what we need the computer.

  26. lee Says:

    sure you not got it the other way round, the more disk access is better with SSD as thay perform better (ignoreing the cost side) then an hdd, more so when it comes to small files and vista does that alot (vista seems made for SSD)

    for norm day to day stuff users not likey need SSD but if you got the money its nice to have one less thing that makes noise and heat and have speed, with all the tests and all if you used an SSD based computer programs and other things just respond faster and open faster

  27. Anon Says:

    Correct me if I’m wrong, but isn’t the disabled cache is helping out the SSDs? According to Tom’s Hardware, the SSDs are comparatively less efficient when reading and writing than when sitting idle, and disabling cache is keeping the drives idle. Under normal use, cache is enabled and users visit different sites, resulting in quite a bit of drive access that doesn’t occur in your test (since the website data is stored in RAM). I would enable cache, but force the browser to fully reload each page (Ctrl+F5) a certain percentage of the time to simulate caching page data to the drive.

    Also, closing and reopening Firefox tests Vista’s ability to cache commonly used applications in RAM more than anything. Which drive was the first to be tested?

  28. Anonymous Says:

    Test seems biased towards SSDs in multiple ways

  29. Pat Says:

    This test doesn’t really use the hd/ssd at all. Repeat this test with your webbrowsing-script, but additionally simulate some HD access. I really mean “some” access! Not continuous access because that would be non-realistic. But barely no access at all isn’t representative either. This could be, like, play 3 minutes MP3 every 10 minutes, open/close/repeat a maybe 5mb big powerpoint presentation/word document, and so on.

  30. Glenn Says:

    “Update: Bar graph reformatted to start at an X axis of 0:00. Previous version of graph showed same exact numbers, but was a “zoom in” that showed from 3:07 to 3:28.”

    YES — thank you! That stuff drives me crazy; see Howard Duff’s classic “How to Lie With Statistics” for more info.

    Most sites wouldn’t have bothered with the change, so THANK YOU for the extra effort!

  31. George Says:

    I would suggest a simple test: a laptop with wireless off, bluetooth off, sound off, and brightness to minimum. Then tun an antivirus check on all files. I do believe that SSDs will do this test quicker and with less power.

  32. Brian Says:

    Tomshardware.com was critized for the same flawed type of SSD articles relating to battery comsumption just a few weeks ago and TOMs re-made another article about SSD power comsumption.

FIND A REVIEW
Laptops
All Product Types Accessories Cars Digital Camcorders Digital Cameras eReaders GPS Laptops MP3 & Video Players Projectors Smartphones Software Storage Tablets / MIDs VoIP Wi-Fi
All Subcategories
All Subcategories All-Purpose Budget Business Desktop Replacement Gaming Multimedia Netbook Nettop Rugged Student Tablet PCs Ultraportable
Brand
Acer Alienware Apple Archos ASUS Averatec BenQ CTL Corp. Dell Digital Storm eMachines Emtec Everex Fujitsu GammaTech Gateway General Dynamics Getac Gigabyte Hercules HP HTC iBuyPower Intel Lenovo MSI Nokia Nvidia OCZ OLPC OQO Origin Panasonic Sager Samsung Sony Sylvania Systemax TabletKiosk Toshiba Verizon Viewsonic Viliv VooDoo Workhorse PC ZT Systems
Minimum Rating
Any Rating Editor's Choice 4.5 Stars 4.0 Stars 3.5 Stars 3.0 Stars
Screen Size
10 11 12 13 14 15 16 17 18 20 4 5 6 7 8 9
Resolution
1024x576 1024x600 1024x768 1200X800 1280 x 720 1280x1024 1280x768 1280x800 1366x678 1366x768 1440x1050 1440x900 1600x768 1600x900 1680x1050 1680x945 1920x1080 1920x1200 800x400 800x480
Weight Range
10.1 - 12.0 pounds 12.1 - 14.0 pounds 14.1 - 16.0 pounds 2 lbs 2 pounds and under 2+ lbs 2.1 - 4.0 pounds 4.1 - 6.0 pounds 6.1 - 8.0 pounds 8.1 - 10.0 pounds Over 16 pounds Under 2 pounds
more options
SUBSCRIBE