• On BNET: 3 worst things about the iPhone 3G S
August 27, 2007 5:02 AM PDT

OLPC battery life: What's the real story?

by Peter Glaskowsky

On Sunday night, 60 Minutes reran a segment on the OLPC (One Laptop Per Child) project originally broadcast in May. I missed it the first time and never got around to watching it online (in Windows video format here, for example; there's a full transcript on olpctalks.com).

Hearing OLPC representative Walter Bender repeat the claim of "10 or 12 hours" of battery life "with heavy use" reminded me of an open question from the last few times I blogged about the OLPC project. What is the battery life of this machine, really?

Back on July 21, the News page for the OLPC wiki (archived here) said this:

We ran battery-life tests on 75 B4 laptops: results will be reported on the OLPC wiki this week.

But I looked around on the wiki during that week and didn't see anything.

So while Andy Rooney was blathering on about kitchen gadgets he doesn't use, I started looking around again, and I finally found results. It was a roundabout search. The OLPC wiki's own search engine can't look for multiword strings, and Googling the site produced no results for "battery testing," but "battery test" gave me a page of test results without any explanation.

That led me to a bug report on the OLPC developer site and an explanation of the results back on the wiki. Whew.

As it turns out, battery life is not very good. The best of the NiMH (nickel-metal hydride) batteries produced a little over 4 hours of operation. Of the two brands of lithium-ion batteries tested, one was about the same as the NiMH batteries; the other ran for a little over 5 hours. (Some of the NiMH batteries died after only a few minutes, but the bug report explains this away as a firmware problem.)

And this was with the machine idling--running nothing at all. Under "heavy use" it's likely the battery life will be less than two-thirds of these figures; maybe half or less.

I am certain that battery life will improve over time as the software and hardware are tuned, but these numbers make it pretty clear that the claim of "10 to 12 hours" is no longer within reach. That's disappointing. To me, this means that much of the long-term promise of the OLPC project can't be achieved until the next major revision of the platform.

I've sent an e-mail to the OLPC project asking for updated information, and I'll pass along whatever I get here on Speeds and Feeds.

Originally posted at Speeds and feeds
Peter N. Glaskowsky is a technology analyst for The Envisioneering Group. He is a member of the CNET Blog Network, and is not an employee of CNET. Disclosure.
advertisement
Click here!
Recent posts from News Blog
Neil Young Archives Blu-ray: Rip off?
Acronis revises survey results about backup habits
Acronis miscalculates data on users' bad backup habits
Flickr co-founder presses beta button
Comcast, Sony open retail store
Cox to try coaxing the Internet into submission
Was InfoWorld's CTO of the Year award a year late?
VMWare VI4 renamed to vSphere
Add a Comment (Log in or register) (7 Comments)
  • prev
  • 1
  • next
There is a black and white low power mode..
by Charbax August 27, 2007 9:47 AM PDT
And I think they only recently started optimizing the DCON process that shuts off the main CPU most of the time when it's not needed, saving a lot of battery time.

The black and white mode reflective mode consumes much less power, so the numbers might be in such e-book mode when DCON will be kicking in to put the main CPU on stand-by mode.
Reply to this comment
Good points
by Peter N. Glaskowsky August 27, 2007 10:40 AM PDT
If the CPU is running without benefit of power management-- like PCs often do before OS-based power management kicks in-- then perhaps these test systems are consuming a couple more watts than they ought to. But this also means the OLPC must be a long way from being production ready.

The backlight issue is another good point, since the tests were performed with the backlight on-- but the most common operating mode of the system will be indoors with the backlight to get a color display, so I'd have to count this as "heavy use" in the sense of the original 10-12 hour claim. Reading a book outdoors in sunlight isn't heavy use by any standard (and I doubt anyone's going to spend even 4 hours a day doing that anyhow).

. png
View all 2 replies
Ebook mode lifetime - > 13 hours.
by Jim Gettys August 27, 2007 10:39 AM PDT
Yes, we are not finished with power management. There are several major power savings yet to be had, when running in typical operation.

The major feature we have is the ability to read books (screen on, but processor suspended). In this mode with the screen on (something unlikely to be the case for any human I've ever seen). I have measured this at over 13 hours, occasionally flipping pages (the system wakes up extremely fast).
Reply to this comment
Thanks for the info
by Peter N. Glaskowsky August 27, 2007 2:43 PM PDT
I also appreciate all the additional data you provided in email, and I'll post a follow-up soon.

. png
(7 Comments)
  • prev
  • 1
  • next
advertisement

Can RIM get its mojo back?

The new BlackBerry Tour, carried by Verizon and Sprint, arrives Sunday, even as RIM seems to be losing sales to exclusive devices like the iPhone and Pre.

With Chrome, Google reignites the OS wars

roundup Google Chrome OS, due in 2010, underscores the Web giant's cloud-computing ambitions and opens new competition with Microsoft.
• What Chrome OS has on Windows that Linux doesn't

About News Blog

Recent posts on technology, trends, and more.

Add this feed to your online news reader

advertisement
advertisement

Inside CNET News

Scroll Left Scroll Right