Bennett Ring speaks to Neal Robison, Senior Director of Global Software Alliances, about the impact AMD’s unique CPU and GPU combo powering the next-gen consoles will have on the company place in PC gaming.
PCPP: Let’s get the bad stuff out of the way first – can you give us an update on AMD’s stuttering issue?
Robison: The latest release of our drivers has addressed most of the issues that were brought to our attention. We spent time with engineering resources and feel we’ve gotten a grip on most of the issues many users have posted.
So totally fixed then?
Yeah, Absolutely. This was the focus of the last major driver release.
Glad we could get that out of the way. Onto the interesting stuff about your work with next-gen consoles and how it’ll impact PC gaming! With both PlayStation 4 and Xbox One using AMD’s Graphics Core Next (GCN) architecture, how will that impact performance on ports that feature on both consoles and PC?
I’ve been involved in game dev and the gaming industry for over twenty years, and this is the first time that I’ve ever seen this much alignment on architecture with the consoles and PC. For us, it signals a huge benefit that developers can really spend the time to create the content on the x86 architecture, and that’s going to work across both consoles and PC. Obviously, there are going to be optimisations for each one of those platforms, but overall this will help reduce a lot of time and cost in game dev that’s done today.
In terms of performance, though, do you think this will give your PC parts a benefit over NVIDIA in cross-platform games, and if so, what are the strengths of the GCN that NVIDIA can’t replicate?
You’re talking about the very basics of the game being coded to take advantage of the GCN architecture. If you start from that very base layer, if you have that same architecture in a PC, you’re going to get the best experience possible. We were chosen because the AMD GCN architecture was seen to be superior, giving better flexibility especially thanks to heterogeneous compute, using the graphics architecture for computation beyond standard graphics.
So do you believe the optimisations that are unique to GCN, won’t be able to be replicated by NVIDIA?
It makes it really difficult. NVIDIA’s still going to do a good job with PC graphics, certainly they’ve built up a good reputation in many areas. But I think overall the fantastic performance you’re going to see in these games, NVIDIA won’t be able to touch those unique optimisations. If they do, it’ll come at the performance of their cards.
We know both consoles are using GPUs based on GCN architecture, but are they identical in their designs, in the same way a Radeon HD 7950 is simply a cut-down 7970? We know that the Xbox One has 50% fewer cores than the PlayStation 4, but in terms of the actual layout of the GPU, is it the same method as PC graphics cards, where one is a pared down version of the other?
They’re actually unique designs, but they come from the same architecture. I can’t go into any more detail as my friends at Sony and Microsoft will get upset with me. But overall, they’re derived from the exact same architecture.
AMD has had eight-core CPUs on the market for a few years now, and you were a few years too early as nobody was coding for so many cores. Now that we are going to see eight cores in both consoles, how will this impact the use of eight cores on the PC? Do you think this is going to turn a corner in the next year or two, where it becomes the norm?
You’ve hit the nail on the head. We’ve seen that many PC developers have embraced multithreaded coding in their games, but this opens it up to a much broader audience. Because these types of cores are now available in the consoles, it’ll give the developers a lot more security and confidence that all the work they’re putting into multithreading is going to carry over and mean something for traditional PC gaming audience. These cores have been around for a long time, we’re seeing other software developers starting to take advantage of the threading. The clock speeds of these chips haven’t increased a great amount over the last few generations, but you’re starting to see greater use of multithreading to improve performance.
How much harder is it to code for eight cores? We’ve had four cores for years and yet we still see brand new games coming out that rely on just one or two cores. Is the switch going to take 12 months? Two years? Five years?
I think it’s a much shorter term. If you look at multithreaded, it’s been around for a while. The PS3 even had lots of cores that had to be juggled. This isn’t new for anybody. But what you’re seeing now is something that hasn’t happened before, that is a very uniform approach across the platforms that the publishers and developers care about. Therefore it’s a lot less risky to explore and expand into multithreading than ever before.
At this year’s E3, many keen-eyed PC gamers noticed that the next-gen demos on Xbox One and PlayStation 4 didn’t look as good as the PC demos. This is a new phenomenon. In the past, the new consoles always looked as good as the PC, if not better, at launch. Yet it feels like this generation the PC will stay out in front of the new consoles. Why do you think this is?
I think it depends on the games you looked at. It was still very early with the tools and environment for a lot of those games. I think if you looked at the lectures given by Mark Cerny, a lot of the things that Microsoft and Sony are doing to customise the performance out of these machines, I think you’re going to see some fantastic looking games. At the same time, the PC has always been on the front. With the increase in performance over the last couple of years, we’re literally at a breakneck pace, delivering more and more graphics performance for PC game devs. I think that’s going to continue to be the case. The PC will always be out in front, because the cost of the part, and the type of the part we supply, are much more expensive, bigger and hotter than anything you’re going to see in a console. Games are going to look great on both sides, but if you look at pure visual fidelity, the PC is able to display on multiple screens, driving a tremendous number of pixels on the PC, compared to the consoles which use a single 720p/1080p screen.
What’s the most exciting thing for you happening at AMD in the rest of 2013?
For me personally, the work we’re doing with DICE on Battlefield 4. This game is unique in all of the PC game market in terms of visual fidelity. Our team have been very deeply engaged with them on optimisations. In fact they used our hardware to debut the game at E3. 64 people all playing on AMD hardware – all running Radeon HD 7990s. Frostbite 3 is one of those few engines that can harness all of the horsepower that the 7990 has to offer. It’s going to look great on other graphics cards, but for that ultimate experience the 7990 was the one. So optimising games like that is really exciting. At the same time, working with our other partners at Sony and Microsoft to make sure that all game devs across all platforms can harness the power of the GCN architecture and AMD technology behind it is one of the things we love about being about this industry.