When Advanced Micro Devices created its own stand-alone graphics division, Radeon Technologies Group, and crafted a new brand, Polaris, for its upcoming graphics architecture, it was an admission of sorts. AMD championed the combination of processors and graphics into a single chip, dubbed the accelerated processing unit (APU). But the pendulum swung a little too far in that direction, away from stand-alone graphics. And now it’s Raja Koduri’s job to compensate for that.
I interviewed Koduri at the 2016 International CES, the big tech trade show in Las Vegas last week. He acknowledged that AMD intends to put graphics back in the center. And he said that 2016 will be a very big year for the company as it introduces its advanced FinFET manufacturing technology, which will result in much better performance per watt — or graphics that won’t melt your computer. Koduri believes this technology will help AMD beat rivals such as Nvidia. AMD’s new graphics chips will hit during the middle of 2016, Koduri said.
Beyond 2016, Koduri believes that graphics are going to get more and more amazing. Virtual reality is debuting, but we won’t be completely satisfied with the imagery until we get 3D graphics that can support 16K screens, or at least 16 times more pixels on a screen that we have available on most TVs today. Koduri wants to pump those pixels at you at a rate of 240 hertz, or changing the pixels at a rate of 240 times per second. Only then will you really experience true immersion that you won’t be able to tell apart from the real world. He calls it “mirror-like” graphics. That’s pretty far out thinking.
Here’s an edited transcript of our conversation.
VentureBeat: Why did you create a stand-alone graphics division within AMD?
Raja Koduri: The context of creating the stand-alone group is quite simple. Our vision is instinctive computing. We want a focused set of people inside the company, almost in a startup mode, that think and do things around instinctive computing. That’s where the big growth opportunities are. We believe we have an engine to make that work. We’re only one of two companies that can do this.
We haven’t been focused on discrete graphics over the last several years. We’ve pursued integration and other strategies. But we believe that to get to an instinctive computing world, discrete graphics will be central. It’s a very healthy market, 80 million discrete APUs sold last year. We believe that will grow, not shrink, because of the trends we’ve seen, like VR. We want a focus from business, engineering, architecture, content, all under one division.
VentureBeat: Getting the APU established and accepted was the first job. Now you’re swinging the pendulum back a little?
Koduri: Remember a few years ago when we talked about the Future is Fusion? Now we’re back to the Future is Fusion again, because of the performance requirements, the stability requirements. We organized this idea called fast pixels, deep pixels, and immersive pixels.
You’ve been watching this for a long time. If you look at the pixel rate, 640 by 480 was the key gaming resolution 20 years ago. Then 1024 by 768, then 720p, 1080p, Ultra HD (4K), and 5K panels. If you look how fast we’re gaining pixels, and forget about how complex the pixels are, we’ve gone from some really low rates to almost a gigapixel. But this is just at 60Hz. We are getting beyond that. One of the things that we’re excited about in the work we’re doing with VR. [Yet even] this image is substantially lower quality than a simple hand mirror that you look at. A mirror has much better contrast, much better pixel resolution, infinite pixel resolution.
Our goal, the path we’re on, is to get to that mirror-like quality. What the eye can see in a display today is a small fraction of what it can see in the outside world. That’s the next set of technologies that we and our partners at the display manufacturers are bringing in. You look at it and think, “Oh my God. This is so much better.”
VentureBeat: You’re very excited about VR?
Koduri: We’re just entering the VR era. You see all these VR headsets. If you see what it takes to drive a VR headset, the pixel rate requirements are almost doubled up compared to the previous iteration. That’s driving up demand for discrete graphics quite a bit. Now, if you push that forward, when you get to 16K by 16K resolution, 120Hz, you get to a pixel rate of 6 billion pixels per second. We’re not going to get there if we just rely on Moore’s Law. We have to do disruptive things to get there. That’s the goal of our division, to get to the immersive era. We need to double up our products and technology, step by step. You’ll see the key initiatives and technology this year. You’ll see more next year.
There are several opportunities to take us to the immersive era. We’ll be working with game developers and engine developers and so on. If we keep on the current trajectory, we need a million [uncertain – 6:12] per year to get us to the immersive era. This includes the performance you need not just at 200 watts. We need this performance at five watts, so that the VR experience is completely mobile. You’ll need that sense of presence.
When I set the goal, I said, “We need to get here in our lifetime.” We can’t do that with Moore’s law and hardware alone. We have to unleash software on this problem. We’ve been working with developers on all of these ideas. How can we get 16K by 16K displays refreshing at 240Hz with the picture that you want to draw? Developers want more control, on their side. They want console level GPU access on the PC.
What they’ve been able to achieve on consoles in the current generation, versus the current high-end PC — the current high-end PC specs are at least four to eight times faster than current consoles. The Fury X is an eight teraflop machine. The PS4 is a two teraflop machine. It’s four times more compute in that single Fury. You can build a dual Fury PC. But PC doesn’t give you that much better an experience with cutting edge content, because they can extract more performance from a console. They’re also investing a lot of IP into that architecture. They’re doing some really clever things that are not possible on the PC yet.
VentureBeat: What’s the solution?
Koduri: The solution is, we’re opening up the GPU. It’s a point in our journey as far as what we’ve been doing with software over the last couple of years. We’re giving direct access, unprecedented control of the GPU, to the developer. Completely open source software. This has never been done on GPUs. It’s been done everywhere else, but not on GPUs. A large collection of tools, libraries, and SDKs, all completely open source and available to developers.
We’re inviting participation in this effort from both our competitors and our partners. We’re getting some really good traction there. You’ll see a lot more. GDC will be a key event. You’ll hear a lot more about GPU Open and some of the partners that we’re signing up to be a part of it. That’s on the software front. I wanted to highlight that, because unlocking the potential for immersive and instinctive computing is all about software. We believe software is the key.
Our competition’s strategy has been proprietary software and locked down boards. That’s not going to increase the capability of the GPU. We want to increase the health of the GPU and take it far beyond just the PC gamer, the PC enthusiast. That’s the sole reason for us to do this big GPU Open initiative.
We have some exciting hardware announcements as well. This is designed for FinFET. Our guiding principle for the Polaris architecture was power efficiency. We have the new naming scheme for our architectures. It’ll be based on galaxies, star systems, and stars. You’ll see more of this coming in the future. Polaris is the beginning of our journey through space.
VB: Does the Polaris brand supplant the Radeon brand?
Koduri: It’s an architecture codename. It’ll still be Radeon something something on the box. But we didn’t have a consistent architecture name like our competitors do. It was hard, because for people, including yourselves and some of the press and enthusiasts—This family of chips has this architecture and a similar class of features. You can group them easily together.
When we set to design this GPU, we set a completely different goal than for the usual way the PC road maps go. Those are driven by, the benchmark score this year is X. Next year we need to target 20 percent better at this cost and this power. We decided to do something exciting with this GPU. Let’s spike it so we can accomplish something we hadn’t accomplished before.
The target we set was to do console-class gaming on a thin and light notebook. What does that take for the GPU in terms of power and configuration? I’m proud to say we’ve accomplished that goal with this GPU.
VB: Is that with a generation coming in 2016?
Koduri: Yes. We have two versions of these FinFET GPUs. Both are extremely power efficient. This is Polaris 10 and that’s Polaris 11. In terms of what we’ve done at the high level, it’s our most revolutionary jump in performance so far. We’ve redesigned many blocks in our cores. We’ve redesigned the main processor, a new geometry processor, a completely new fourth-generation Graphics Core Next with a very high increase in performance. We have new multimedia cores, a new display engine.
This is very early silicon, by the way. We have much more performance optimization to do in the coming months. But even in this early silicon, we’re seeing numbers versus the best class on the competition running at a heavy workload, like Star Wars—The competing system consumes 140 watts. This is 86 watts. We believe we’re several months ahead of this transition, especially for the notebook and the mainstream market. The competition is talking about chips for cars and stuff, but not the mainstream market.
In summary, it’s fourth generation Graphics Core Next. HDMI 2.0. It supports all the new 4K displays and TVs coming out with just plug and play. It supports DisplayPort 1.3, the latest specification. It’s very exciting 4K support. We can do HEVC encode and decode at 4K on this chip. It’ll be great for game streaming at high resolution, which gamers absolutely love. It takes no cycles away from games. You can record gameplay and still have an awesome frame rate. It’ll be available in mid-2016.
VB: In years past people looked to something like movies for inspiration. They’d say they wanted to do The Matrix on a game console. Do you have something that communicates to people what you mean when you say you can give them better graphics? Also, there are so many advances here — I wonder if you believe that’s enough to justify a new console generation. Microsoft upgraded from 720p to 1080p within the same console on new generations of the same chips. That wasn’t a reason to do a new console. But if there is going to be new consoles, what do you think are some things that would justify that?
Koduri: The answer is encoded in the pixel rate charts I’ve shown you. The jump from 1080p to 4K is much larger. VR’s needs are also much larger. There’s demand for more pixel throughput and more GPU processing power. That, to me, is a compelling reason. The pixel counts are quadrupling. You can buy a 4K TV for under $300 now.
VB: VR is coming and it’s something people will probably want. But what vision do they want to get to with that?
Koduri: I go back to this picture. This is what’s driving us. We’re just entering the VR generation now. We’re here. Even this requires much higher throughput than today’s graphics to get there. Part of the Polaris stuff is making the VR transition be a fruitful thing for even thin and light notebooks.
Today you look at what Oculus and everyone else is talking about as far as specs. It’s still limited to high end desktop PCs. With Polaris we want to bring that down to a much larger part of the market. But you see where we’re aspiring to be in the coming years. Our realization is that to get to that next step, it’s not just hardware evolution. In fact, most of the steps up to this VR point have been enabled, if you look at it, by significantly faster hardware at every step. The software pipeline hasn’t changed too much after shaders were introduced.
Fundamentally, when shaders were introduced, we could calculate light in a programmable way. All that happened, if there was a significant thing between then and now—Shaders were introduced in 2002. Between 2002 and 2015, we took a lot of this technology as an industry and took it to all platforms, to smartphones and tablets and stuff. On PC we just drew the resolution up higher and higher.
Now the next step is to get to the goal of photorealism in VR, to movie quality. That’s now achieved by rendering a frame for several hours. The realtime guys are going to get there in a much more efficient fashion. We’re working on that together with the software industry.
One key thing that’s different about this compared to our previous setup is that content is a central part of our strategy. It’s the highest priority, which is significantly different. Since last January, this time last year, we projected and predicted that we’d be huge. The number of people I myself employ — I thought it was going to take five years. But now there’s so many companies and so much content being developed on VR. A lot of it is evangelism by AMD, that we have to take VR seriously.
We also took VR outside gaming. If you saw the piece we did with the Smithsonian–We got a lot of startups, a lot of companies doing VR for education, content and modules for education. That’s one of the significant differences. Content is king for us.
VB: Your organization is becoming more independent. Are there any more changes you’re looking forward to that solidify that?
Koduri: We still have a fair amount of work to do to accomplish all the goals on the content side and on the business side. We’re constantly optimizing the organization and bringing in new talent, both internal and external. We brought in Scott Wasson to bring an enthusiast gamer perspective to the company. You’ll see acquisitions like that where we increase our connections directly more and more.
Another thing I should mention, beyond the focus on gaming — we have substantial plans around our workstation business and what we’re doing for the CAD/CAM side of the world, the movies, other content creation. We have an excellent pipeline there, especially as Polaris-based workstations come in this year. We expect to make a run at the workstation market much more aggressively, with much more compelling hardware and software than any time in the last 10 years. That’s another area where you’ll see a lot more work.
We’re also doing some very interesting work on VR cinema. That in itself could be an interesting piece, the world of VR cinema and what’s happening in Hollywood as far as content production. All sorts of companies in L.A. are doing interesting work.
VB: From the startup perspective, we’re seeing a generation of new funds being created. People who invested tens or hundreds of millions in mobile game companies are now setting up funds to invest in VR and eSports. They’re leaving mobile gaming behind. They’re investing in all sorts of things, from sensors to VR to esports.
Koduri: PC-based eSports? eSports is known as being less visually demanding than, say, a first-person shooter game. Does eSports VR become — right now the popular wisdom is that you need a 290X or a 970 or higher to run a VR experience. Does the rise of VR and eSports bring it down a notch, because you have less visually demanding games to deliver a VR experience with a smaller technology footprint?
As part of our Liquid VR strategy, we talked about the sensor stuff and strategies around that. That’s definitely part of our strategy, sensor computing and getting all kinds of sensors connected for full presence. You’re absolutely right. It’s very challenging, but we’re doing some amazing stuff on the VR front, and a lot of sensors, too, the cameras. There’s a huge opportunity there.