The Apple iPad 2 Review
by Brian Klug, Anand Lal Shimpi & Vivek Gowri on March 19, 2011 8:01 PM ESTBattery Life
Testing battery life on a tablet is pretty grueling as these things are effectively smartphones with gigantic batteries. What would last for only a few hours on a smartphone can easily last over ten on a tablet. The iPad 2 is no exception.
As iFixit discovered, the iPad 2 features a nearly identical battery to the original iPad (25Wh vs. 24.8Wh). We already established that display brightness hasn't gone up so the only real impact on battery life will be because of the A5 SoC.
The iPad 2 WiFi's battery, courtesy iFixit
The ARM Cortex A9 tends to be more power efficient than its predecessor, the A8. With a shallower pipeline and the ability to execute instructions out of program order you get better performance per watt in most cases. As a result the A5 should complete bursts of instructions quicker than the A4 and get back to sleep sooner.
To put this to the test we put together a battery life test that combined our non-Flash smartphone web browsing test, checking for email every 15 minutes (via a Gmail account receiving ~100 emails per day) and constant local music playback while all of this is going on. The results are pretty much as expected:
Over WiFi the Motorola Xoom and iPad 1 deliver very similar results. The iPad 2 pulls ahead by around 8%, it's not a huge advantage but it's there. Over 3G there's a noticeable drop in battery life, putting both of the iPad 2s below 10 hours and kicking the Xoom down to under 9 hours. All of the numbers on this chart are respectable however and you can expect a good day's worth of use out of any of these tablets.
Note that while the iPad 2 did better than its predecessor if you were to run an app that pegged both CPU cores at 100% you'd see reduced battery life on the iPad 2 vs. the iPad 1. Granted you'd also see better performance on the iPad 2.
Video playback is obviously an important function of these new tablets so we put together a H.264 decode playback test as well. For this test we transcoded a copy of Quantum of Solace from a 1080p BD source down to a 720p H.264 using Handbrake's Normal encode profile with one modification. In order to enable smooth playback on the Motorola Xoom we had to disable b-frames in the encode. The results are below:
The iPad 2 consistently lasted longer than the original iPad in our video playback test. The Xoom however did noticeably worse. Unfortunately only a portion of the H.264 decode pipeline is accelerated by NVIDIA's Tegra 2 in this test. When playing back anything more strenuous, portions of the pipeline are handled by the CPU cores in software. If you were to enable b-frames the Xoom would not only stutter but its battery life would drop to around 6 hours.
On the iPad 2 its Cortex A9s are mostly powered down during this test, on the Xoom the A9s are helping with the heavy lifting for video decode and as a result we see lower battery life. Kal-El is expected to address this issue head on.
Facebook
Twitter
RSS
189 Comments
View All Comments
PeteH - Saturday, March 19, 2011 - link
In the Garage Band section:"There are three Smart Instruments - Piano, Bass, Guitar, and Drums."
I'm pretty sure that "three" should be a "four."
VivekGowri - Sunday, March 20, 2011 - link
Ahaha, I'm an idiot - thanks for catching that, it'll be fixed.PeteH - Sunday, March 20, 2011 - link
As far as typos go that one isn't remotely bad. I once published a spec (internally) that had a section detailing how asynchronous boundaries were handled in my section of a chip. Unfortunately I had titled that section "Cock Domain Crossings."Anand Lal Shimpi - Sunday, March 20, 2011 - link
A few years ago I used the word overcocking instead of overclocking in an article.UNLK A6 - Saturday, March 19, 2011 - link
I'd like some clarification about LINPACK and Geekbench. Are these benchmarks created by compiling some portable code for each platform as a measure of floating point performance? Or, is this supposed to be some measure of how fast one can do linear algebra or DSP on the platform? On Mac OS and iOS, one wouldn't compile say LINPACK for this but use the hand-tuned LAPACK/BLAS and DSP routines built into Apple's Accelerate Framework. The difference between the two can be huge. Which do these benchmarks purport to supply--generic floating point performance or available linear algebra and DSP performance on the platform?metafor - Sunday, March 20, 2011 - link
I believe Linpack on both iOS and Android are plainly compiled (by the JIT in the case of Android) to run on the platform. They don't make any calls against the onboard DSP's nor do they use NEON beyond what the compiler is able to auto-vectorize.name99 - Sunday, March 20, 2011 - link
Apple supplies all the Linpack routines in optimized NEON code as part of the OS (in the Accelerate framework). Intelligent apps that need them use those routines.Android, as far as I know, does not provide an equivalent.
You can use apps that deliberately bypass these iOS routines if you wish to get a handle on the raw FP performance of the hardware, but
(a) it doesn't give actual linear algebra performance, if that is something your app or algorithm really cares about AND
(b) it's kinda dumb because if you care about fp performance in any way, you'll be using NEON, so what's the value in a benchmark that doesn't exercise NEON?
nimus - Sunday, March 20, 2011 - link
I hope AnandTech can do a comprehensive comparison of the usability/feature strengths between the Android, Apple iOS, BlackBerry Tablet OS (QNX), HP webOS, and any others tablet OSes.It will be interesting to see how the Windows Tablet OS will be able to compete when it finally is released for ARM processors.
KidneyBean - Sunday, March 20, 2011 - link
I'm using a tablet, so I can't see the mouse-over pics :-(tcool93 - Sunday, March 20, 2011 - link
I don't know where the reviewer gets the idea Netbooks are much faster. That is nonsense. Here is a video showing an ARM 9 processor being just as fast, yet the ARM 9 processor is running 1/3 the speed of the Netbook Atom. (500mhz vs. 1600mhz for the Netbook).http://www.youtube.com/watch?v=W4W6lVQl3QA&fea...
The Netbook also has a graphics accelerator in it, and the ARM shown in this video doesn't.