A CPU Bound Case: Streaming YouTube Videos

Bandwidth enables new usage models, it's what we saw on the desktop and it's what we're seeing in the mobile phone market as well.  Streaming video to mobile phones is a definite bandwidth consumer, and something that's truly enabled by 3G wireless technologies - but at what cost to battery life?

On the Blackjack we used the YouTube Mobile site (m.youtube.com) with a hack to enable playback on Windows Mobile 5.  We played back the same video over and over again until the battery died.  Unfortunately the longest video we could find on YouTube Mobile that was stressful enough for our test measured around 5.5 minutes, meaning that by the end of the 3+ hour process we were ready to die. 

Things were a bit easier on the iPhone as its YouTube implementation has a larger selection of videos, including our test video which was over 40 minutes long, thus requiring fewer clicks on our part to keep the video looping. 

Our YouTube test is a little misleading when it comes to the iPhone because there's more than one variable being changed here.  When switching between Wi-Fi and EDGE on the iPhone, the YouTube client automatically requests a different quality video depending on what wireless connection you're using.  The stream over Wi-Fi looks absolutely beautiful, while over EDGE you get a much lower bitrate stream that's more reminiscent of what YouTube looks like on the web.  The problem with this is that we're unable to measure the impact of Wi-Fi vs. EDGE alone, as the flipped results show (EDGE outlasting Wi-Fi). 

Looking at the Blackjack results helps us make sense of the iPhone's behavior.  Streaming video from YouTube is predominantly CPU bound, so much that the primary determinant of battery life is the CPU itself, not driving the wireless antennas.  The end result is that while 3G draws more power than EDGE, the difference is around 6% because the CPU is eating the vast majority of the power budget and its role is identical regardless of whether the phone is running in 3G or EDGE mode. 

If we assume that the same CPU bound limitation exists on the iPhone, the fact that while decoding lower quality video the iPhone would last longer makes sense.  Lower quality video means a much lighter load on the iPhone's CPU, thus resulting in longer overall battery life.  In this case, the significant reduction in quality results in an equally significant increase in battery life (37% or close to an extra 1.5 hours). 

Still an Impact while Mostly Idle? The Email Test The Real Shocker: Impact of 3G on Talk Time
Comments Locked

36 Comments

View All Comments

  • Tegeril - Friday, July 13, 2007 - link

    No, because EDGE is disabled for data during voice calls and Wifi remains on.
  • mino - Tuesday, July 17, 2007 - link

    Well, explain that to me how was I talking with a friend and browsing DT at the same time with my nokia now ?

    Even the good old GPRS allows for simultanious data/voice scenario, that is actually one of the points of the packet-based technologies.

    If you have standard 4+2 GPRS/EDGE timeslots available you just have the bandwith during call reduced to 3+1 scenario as voice requires 1+1 slots for its operation.

    The other thing is if you have phone bad enough that manufacturer disabled/did not implement this functionality...

    Even my P800 form 2003 coud do that FYI..
  • DavenJ - Friday, July 13, 2007 - link

    No he's right it should say EDGE. Phone calls can only come over the EDGE network and not through Wifi. This is a measure of talk time which would mean a phone call. It would be nice, however, do have VoIP through Wifi for those cases where your either roaming or in a dead spot.
  • michael2k - Saturday, July 14, 2007 - link

    It's not a typo, you misunderstand the test.

    It's talk time with a broadband connection turned on. You can, theoretically, turn wifi off during talk to extend the iPhone life, but for a comparison between 3G and WiFi battery life they were talking with, respectively, 3G and WiFi on.
  • VooDooAddict - Friday, July 13, 2007 - link

    WiFi for voice would be very nice. They could even charge me the same minute fees/bucket. There are plenty of homes with very good wifi coverage but crappy wireless cell coverage... works almost like a signal booster.
  • Ataraxia - Friday, July 13, 2007 - link

    I'm pretty sure you guys are all wrong.

    The iPHone has just enough features so people will go out and buy it in rabid frenzy, while the next iteration (in probably 6 months) containing 3G, Thinner/Lighter, etc will push people like ME to buy it (The educated consumer, who really does know how backwards the innards of an iPhone actually is), plus everyone who bought the first generation will dump theirs for the much faster data transfers and buy a second gen iPhone.

    It's all about business and the $$
  • BigLan - Friday, July 13, 2007 - link

    I wonder if the iphone youtube results are so good because the device buffers the video. It could be that you only download the file once, then play it back from cache. Could you maybe repeat the test with different clips every time to see if this is the case?

    Also, are you saying that youtube clips on iphone/wifi look better than on a desktop?
  • DaveLessnau - Friday, July 13, 2007 - link

    If Apple simply went with a user-replaceable battery, most of this problem (not the space issue) would go away. I'm not in the market for a cell phone, but that one point (the battery) drops the iPhone out of the competition for me.
  • Shintai - Friday, July 13, 2007 - link

    Why would anyone, atleast outside the US mobile network, get an iPhone? Its 3-4 years behind atleast.

    Take a cheaper nokia n95. You get gps, video calls (DVD quality, 30fps), TV, msn, 5Mpixel cam, MMS, 3Mbit internet, longer battery life and so on. On a 3.5G network for less than an iPhone.

    The iPhone is like buying a 800Mhz P3 in a fancy design today. Inferiour products for extreme prices. The last sucker aint born yet.
  • sxr7171 - Monday, July 16, 2007 - link

    N95 is not cheaper buddy. I dropped $750 on mine.

Log in

Don't have an account? Sign up now