Digital Foundry vs. iPad 2

While most reviews of Apple's iPad 2 dismiss the revised hardware as an accomplished, if underwhelming successor to the original, a technical breakdown of the new A5 processor reveals something quite different - an unprecedented leap in mobile games-playing power that could conceivably herald the arrival of a new home games console.

This is a hardware design that is about so much more than just a revised form factor and a bunch of cameras. With iPad 2, Apple is laying the groundwork for the future of its gaming business.

As evidenced by these Anandtech benchmarks, iPad 2's raw gaming performance represents anything from a 4x to 7x leap over what was seen in the original version of the tablet, redefining the current state-of-the-art for mobile gaming tech - at least until the launch of Sony NGP.

In this article we'll be looking at how that raw power has been utilised thus far by developers and answering the question of whether iPad 2 is a worthwhile upgrade for gamers over the original. From there, we'll be assembling a muscular body of evidence that suggests that the new A5 platform is powerful and scalable enough to form the basis for a new home console. With Apple's WWDC conference kicking off simultaneously with this year's E3, perhaps Nintendo's Project Cafe will not be the only new gaming hardware announced?

To begin with, let's tackle the basics. Away from the impressive raw specs, what does the extra power of iPad 2 add to iOS's gaming credentials in the here and now? Is there enough software out there to justify an upgrade if you're already using an iPhone 3GS or an existing A4 product such as iPad or iPhone 4?

One of the best technical workouts released so far is Epic Citadel - a bespoke playable demo Epic Games released in order to showcase Unreal Engine 3 operating on the iOS platform. At the time it was lauded as a new standard bearer for mobile graphics, but as the demo was picked apart it became clear that despite the superb visual quality, performance was best described as "variable". This was especially evident on iPad, where the same A4 processor needs to support a significantly higher resolution than iPhone 4 - 1024x768 versus the Retina display's 960x640.

Due to the fill-rate deficit, both Epic Citadel and later Infinity Blade actually ran smoother on iPhone/iPod Touch, so in what ways does iPad 2 redress the balance? Do we see a raw speed increase running code that has seen no new optimisation effort at all? Thanks to the HDMI mirroring function of the new hardware, this is something we can immediately put to the test.

There are a couple of intriguing elements to the analysis. Firstly, iPad 2 blitzes through the "guided tour" parts of the demo with remarkable ease, operating with a fairly consistent frame-rate of 35 frames per second. The fact that we see this level of consistency despite the very different levels of load being placed on the engine strongly suggests that frame-rate is being capped, with the obvious conclusion being that Epic Citadel on iPad 2 could in theory go much, much faster.

The 35FPS cap in itself is also intriguing as it seems like a rather arbitrary figure. As we've also seen recently in our Real Racing 2 tests, the frequency with which frames are output on iOS OpenGL games can be inconsistent with unique and duplicate frames bunched together, making it actually appear less smooth than the kind of locked 30FPS you might be used to from games like Fight Night: Champion and Need for Speed: Hot Pursuit. We see the same thing here, and indeed in many others games - RAGE HD for one.

Checking out John Carmack's latest iOS release offers up some interesting findings too. Unlike Epic Citadel, the game supports TV output on both iPad 1 and its successor, allowing us to put the two tablets up against one another in a head-to-head shootout.

It's difficult to avoid the conclusion that apps which are tightly optimised for the iPad and designed to run at a set performance level will see much benefit at all compared to the same code running on iPad 2. Carmack and his team appear to have hard-coded a frame-rate cap into RAGE that sees performance max out at around 30 frames per second, and as there has apparently been no new update to the game since iPad 2 shipped, it's no surprise that we see the same level of visual quality too.

The only real difference we see in this frame-rate analysis is the very occasional dip in performance on the first iPad, perhaps down to background tasks.

Clearly it's early days for the new iPad hardware, but it's very reasonable to expect software developers to push out new updates that take their existing games and add visual improvements that you'll only see on A5-equipped platforms. After all, the arrival of a new piece of hardware potentially allows them to resell their existing library of games to a whole market, and with A5 set to debut in iPhone 5 and potentially other devices, that pool of potential customers is only going to increase in size as time goes by.