Earlier today, Kuju announced via Develop that it had already started a significant investment in technology for the generation of consoles that follows PS3, 360 and Wii. Codenamed ‘Fabric’ the tech is targeting many-core CPUs. We caught up with technical director Adrian Hawkins to find out more…
What has brought Kuju to be so proactive towards the next generation of consoles?
Simply, we expect that the major hardware manufacturers are set to announce their next platforms over the next year or so – and we want to be ready for it.
The nature of those new devices will be many, many processing cores, representing a real technical leap forward. We think it’s the right thing to do to pre-empt that now, rather than necessarily wait for the widespread availability of dev hardware closer to the consumer releases of the next consoles.
That said, the current generation of consoles will still be strong for some time, and have a competitive second half of their cycle – and our focus now is to make sure Fabric makes the most of those. Many games and developers significantly underutilise the current generation of multi-core hardware, and this is something we have addressed.
Is many-core the main area you will be tackling?
Yes, we’re thinking of platforms that could have processors with cores in the tens or hundreds, potentially.
Historically in this generation many developers complained that they weren’t given enough time to learn current hardware platforms. Are you trying to avoid those kind of things next time around?
Yes. While of course – and this is very Donald Rumsfeld – there are some known unknowns; preparing for many-core devices will structure what we’re doing in the right way to cope with those new machines.
You’ve said the first game will start to use the tech towards the second half of the year. How and when will the rest of the Kuju organisation’s studios start using it?
We’re not going to make an immediate wholesale switch over – but I imagine that by 2010 we’ll have a number of projects in the works using Fabric.
One of the important things about Kuju is that we are a very diverse group. For instance, Zoë Mode makes a number of multi-format music and party games – while Headstrong has recently seen a lot of praise for their action games. Both are successful, but very different. So we’re also planning to address that divergence, too, in the way that we are technically architecting our solution. We believe we have a good solution to cope with those differences – the whole of Fabric is component and level based, so we can chop and change what is needed, while retaining a framework that is going to scale well.
We certainly believe that, over time, as more of our studios move onto the newer generation of platforms, this will be widely used. Of course the timeframe at which that happens will vary.
One of Kuju’s studios exclusively uses Unreal Engine 3, which has become widely adopted and a bit of a de facto standard in the games industry. Does this mean you’ll be abandoning using middleware?
It all depends on the kind of games you are making - Unreal Engine is fantastic for making a number of games. And it’s working well in the studio that uses it – our use of Unreal will continue into the future. We are not going to impose Fabric on all the studios; if a particular technology choice is the right solution for a specific game, then we’ll use it.
This is also a core technology issue – but getting that to work across multiple sites hasn’t worked historically. EA has publicly admitted that its acquisition of Criterion’s Renderware tech was a misstep. And if EA can’t get it right, why does Kuju think it can?
We’re very well set up for multi-studio developments already – a number of our recent titles have actually be developed successfully across multiple sites. In one case specifically, we had five sites working on one game. We have been a multi-site developer for almost a decade now, and have a long history of sharing technology across multiple games.
I think we’ve cracked multi-site work in ways others haven’t, and that we have proven knowledge-sharing between studios. And as an aside, general inter-studio knowledge sharing is something else Kuju has been dedicating itself to, which will help Fabric's success in the long-term.
And from a business development point of view, is it your hope that in theory when the next generation of consoles roll around there could be an agenda-setting launch title made by Kuju using Fabric?
We can’t talk about any plans specifically, but we’d love that to be the case.
Is this development something limited only to Kuju or are you keen to work with other companies to develop Fabric?
Well, this is definitely an internal project – we aren’t looking to sell this or move into the middleware market right now. But part of the reason we are talking about this now is to reach out to and form some new partnerships in technology.
Such as the Intels and AMDs of the world?
Sure, and also the middleware companies – those people that work on particle systems, animation, lighting solutions, physics engines and so on. This is us reaching out to form new partnerships with them and making sure their technology integrates well with ours.
Lastly, isn't it a bit early to be talking about the likes of PS4, the 360's successor and the like?
Not at all - the sooner we as developers can prepare, the better. We think that investing early in this will pay off, not just in the next generation, but in the next half of the current console cycle.