Apple's iPad - The AnandTech Review

Honest Apple: Battery Life

I mentioned this in my iPhone 3GS review and I believe it bares repeating: Apple's battery life estimates have been ridiculously accurate over the past couple of years. I swear Apple must have my office bugged because my battery life numbers almost always equal theirs and they have no access to my test files.

To test battery life on the iPad I ran a modified version of the process I use for smartphones. I use the WiFi connection to browse a series of 20 web pages varying in size, spending 20 seconds on each page (I timed how long it takes me to read a page on Digg and came up with 36 seconds; I standardized on 20 seconds for the test to make things a little more stressful). The test continues to loop until the battery dies. This test is designed to simulate a relatively heavy, but realistic data load on the phone. We're stressing the WiFi radio, SoC, memory and display subsystems here. This should also be the sort of battery life you get when you are using any apps that use data (but not 3D acceleration). The display brightness was set to 50% on the iPad.

That's the test I use for smartphones, but to make it a little more stressful on the iPad I continuously played music using the iPod app, one of the only apps that's allowed to run in the background. I also told the mail app to automatically check my AnandTech email account every 15 minutes. I get a good amount of mail so this constant checking would add another realistic component to the workload. I also ran this same test on ASUS Eee PC 1001P:

The iPad came in just shy of Apple's 10 hour claim. At 9 hours and 45 minutes, it's long enough to get you through the greater part of a day. But calling Apple's battery estimates conservative is misleading. A single charge won't last you all day.

The iPad does last longer than ASUS' 1001P, however ASUS tells us that the 1005PE will buy you another 2 hours of battery life for an extra $79.

What about watching movies? Our resident smartphone expert Brian Klug put together a 6 hour loop of The Bourne Ultimatum. This was from a 1080p source but re-encoded using Handbrake's Normal profile resulting in a 1.3GB 720p rip that was looped three times.

I imported the movie into iTunes, synced it to the iPad and looped the already 6 hour loop until the iPad died. Display brightness was set to 50%, auto brightness control was disabled, automatic email downloading was also disabled.

Video playback is actually a fairly light use case. In these SoCs there's usually a dedicated video decode block, most likely the PowerVR VXD (same as in the iPhone 3GS). This block decodes each frame and sends it to the display output engine. The only things working in this case are parts of the CPU, the video decode engine, memory bus and the display engine. There's no 3D rendering and the vast majority of the CPU is idle. With this in mind, it's no surprise that the iPad can last 13.6 hours when playing back a 720p H.264 movie.

The iPhone 3GS however can almost hit 10 hours performing the same test. This gives you an idea of how much power the display is consuming on the iPad. The netbook doesn't stand a chance by comparison. The iPad was made for watching lots of movies on a single battery charge.

My final battery life test was a 3D gaming benchmark. I ran Real Racing HD on the iPad (and Real Racing on the iPhone) until the devices died. The cellular network was turned off on the iPhone 3GS and brightness was set to 50%.

Anything GPU intensive is going to be the worst case scenario for these SoCs. Here we’re stressing the CPU, GPU and memory subsystem. The added load is reflected in the battery life numbers:

The iPhone 3GS approaches 4 hours, while the iPad manages almost 9 hours while gaming. I’d also consider this to be a best case scenario gaming benchmark. A more intense 3D game could easily drop these numbers even further.