Digital Foundry: iPhone vs. Developers

All of which segues seamlessly into the new kid on the block - the iPhone 3GS. The ARM 1176JZ supplied by Samsung and running at 412MHz is supplanted in the new model by an ARM Cortex A8 running at 600MHz, while the PowerVR MBX chip is replaced with a new, more powerful SGX variant, the importance of which may well extend beyond just the iPhone platform if the recent PSP2 rumours carry any weight. The question is, will the new hardware be used by games and isn't Apple at risk of fragmenting its own market? Is the 3GS in essence far more powerful than it needs to be if developers stick with developing for the base configuration?

"It's not over-specced at all," says Canis Lupus. "Obviously I'm guessing here, but Apple's viewpoint likely goes something like this: 'We want to add video recording to the 3GS. So obviously we need to put a better camera in there. But if the viewfinder on the screen updates laggily, it's going to suck. So we need faster graphics hardware. Also if we boost the CPU, we'll be able to provide some simple video editing without it taking forever... plus everything else will get that bit snappier, like web pages and searching through email. Win!' Remember, they're busy pwning smartphone manufacturers at the moment. If they happen to pwn the PSP or the DS into the bargain, well, that's just a happy accident. That's not going to happen yet, given the well-known money-printing capabilities of the DS, and even the PSP's recent and possibly Monster Hunter-fuelled growth, but still, Apple are building up a pretty big install base..."

"I think developers are just going to use the 3GS chip to enhance what's in the game that'll essentially be the same across all devices," adds Boutros. "Shinier balls in Peggle? Fireworks in Traffic Control... that sort of thing. iPhone dev is pretty risky now, so investing in the smallest share of owners (3GS) is not desirable."

"I think that some companies with a lot of resources (for example, EA) may target both power levels, but most people will leave the 3GS untapped until more users have them," explains Lupus. "Much like with PCs, we may see system requirements creeping up over time... We're planning a couple of projects in advance, one of our later projects will probably start to take advantage of extra 3GS rendering capacity, but as a small company we can't really afford the resources to double-up the rendering architecture to take advantage of pixel shaders etc."

"We think it's an awesome piece of hardware and can't wait to push it to its limits," Firemint's Robert Murray says of the 3GS. "I don't think it will really split the market because I think there are very few developers who are even pushing the limit of the current handsets. The vast majority of games on the platform have very modest performance requirements. If a console developer did this then it would be a different story because most of the games tend to be pushing the limits of the hardware. I think it would be a mistake for Apple to hold back on performance improvements, because they would only invite further competition and the increased performance makes a significant improvement to the usability of the phone overall."

But Murray also believes the incremental upgrades to a base spec iPhone game would work out for leveraging the 3GS in the short term.

"There are a number of different ways we can do this technically. From a design perspective, I think the best approach is to provide enhancements on the 3GS that don't impact the gameplay," he says. "As the 3GS install base grows then eventually it may become the dominant configuration and developers may lead with the 3GS version and port down to other versions, rather than leading with the lower versions and porting up. In the case of a game like Real Racing that pushes the limits, we have to think a bit more carefully, because people expect us to really make the most of all the power available, so we have to consider that in our design."

Speaking of which, Firemint is one of the few developers to have shown what is possible with the 3GS configuration in an exceptional video demo showing Real Racing running on an iPhone 3G, then on the newer model. The difference? Six cars in the original game, and a colossal 40 in the 3GS demo build, running at much the same frame rate, as you can see right here.

Firemint also believes that the 3GS offers around double the rendering power of the original hardware, and is also intrigued by the possibilities of proper programmable shaders and support for the Open GL ES 2. However, doesn't this demo prove that 3GS has considerably more power under the bonnet than has been estimated up until now?

"Increasing the number of cars is an example of where something like a 1.5 or 2 times increase in performance can have a far bigger impact on the results in game," Murray explains. "The car AI, physics and rendering is only one task that consumes cycles, and in general the track rendering is the most significant impact on performance. If we had to draw that track twice then perhaps we could only support 12 cars instead of 40. However, considering that we were already rendering the track and doing all the other 'baseload' things that we had to do on a normal iPhone, 100 per cent of the extra power of the 3GS could be dedicated to rendering cars. So it's not directly proportional."

In a recent interview with Shacknews, id's John Carmack dropped a tech bombshell. He reckons that the same engine that powers the forthcoming PC and console game, Rage, could be shoehorned into the 3GS, despite him finding the USD 99 iPhone price-point being much more exciting for the marketplace as a whole.

"Now I am very excited about what I can do from a hardware and graphics standpoint with the 3GS. With vertex fragment shaders and OpenGL 2.0, I'm pretty convinced that I can actually run the MegaTexture id Tech 5 content creation pipeline on there," he said. "And I'm not sure what game I want to do that with yet, but the combination of seeing people download 700MB files of Myst on there, and the new capabilities, I could do some mind-blowingly cool stuff on there. But that's going to be a much, much smaller market. So there's no way I could justify doing something like Doom Resurrection... but I hope I can wind up doing something in the good old classic way of doing things, where I do something that's just amazingly cool to look at but maybe a little thin on the ground in terms of what the actual substance is there. But do that to kind of get things out and test it and see what's going on."