Advanced Micro Devices is taking a swing at Intel with its new Ryzen desktop processors, and it’s going after Nvidia with its new Radeon Instinct artificial intelligence processors. After playing catch-up for years, Sunnyvale, California-based AMD is back on the offensive.
Lisa Su, CEO of AMD, and Raja Koduri, senior vice president of AMD’s Radeon Technologies Group, hope that their efforts will reinvigorate competition in the computer industry. The Ryzen desktop computer chips, which use AMD’s Zen core designs, will debut in the first quarter. And the new Radeon Instinct machine intelligence chips will debut during the first half of 2017.
AMD says the Ryzen chips are 40 percent faster per clock cycle than the previous generation, and Su is confident that it will beat Intel’s fastest desktop processors on a variety of fronts. And she’s still confident that 2017 will bring great things for the company, and that “AMD is back.”
I sat down with Su and Koduri at an AMD event in Sonoma, California. Here’s an edited transcript of our interview.
VentureBeat: It looked to me like phasing the graphics strategy in here—you’ve taken care of graphics, and now you’re taking care of AI. Is it a kind of one-bite-at-a-time strategy?
Lisa Su: We view graphics as very strategic over the next five to 10 years. When we started with the road map, it was, “Consumer graphics is very important. It’s a core user base for us.” That was Polaris this year. But, no question, our plan was always to compete across the entire graphics end-to-end space. The next phase is compute and making sure we have a very competitive hardware and software platform.
VB: Was that a resource allocation strategy?
Raja Koduri: We definitely have graphics as a focus, a continued focus, a lot of investment on that side. But on the compute side, we had to build up infrastructure. That enormous software stack I showed you, we’ve been working on that. To go to market we needed the right software stack. The timing is now. It all happened to coincide with having the MI25 and the Vega architecture. The pieces are falling together for us in 2017.
We have a lot of work to do in graphics still. We haven’t taken care of it. We have a lot more opportunity, if you look at the size of the market and the dollars that are still there. We have a lot more to grow in graphics alone. Compute is exciting, though. It’s great that we finally have a compelling software stack that goes along with our hardware.
VB: These look like great accomplishments, Zen and Vega. Do you know yet how good they’re going to be for you, though, given the competitive picture? Does this possibly take us back to the days of Opteron, when AMD had a competitive advantage? Or do you have to wait and see where Intel and Nvidia will come in?
Su: Let me comment on Zen and Raja can comment on Vega. We know what the competitive market is like, very well. We believe, whether you’re talking about desktop or servers, there are portions of the market that are not served well by the products that are there today. I absolutely believe we’ll grow share. We’ll see how much we grow as the products roll out. But we feel very good about where we are.
Koduri: We definitely know where we are right now, and the competitive trajectories. We always count on competition executing well. The great thing that you’re seeing in the 2016 momentum, and also going into 2017, is that we’re competing and fighting in every segment. We’re not leaving any segment to the competition. That’s how we think about 2017. We’re going into every segment.
The excitement with Vega is that we’re getting back into the high end of the graphics segment. In 2016 we didn’t have any coverage there.
VB: It looks like you have a very competitive product in the same space as Nvidia. Do you think you’re going to come to care as much about things like self-driving cars as Nvidia does today? Is that a little further down the road?
Koduri: Like I say, we definitely care about all applications of machine intelligence. Our approach is going to be different. We’re not going to take Nvidia’s approach. Our approach will be very focused on key customers and key enabling of those customers. The amazing thing, is that the technology we’re building to scale for the data center, 90 percent of those same things are required for machine intelligence in the auto industry. There’s some additional stuff in terms of qualifications, robustness of the hardware, some certifications and all that we’ll have to do on top of that. But the base infrastructure is there.
We want to go into these markets in a very focused manner. AMD, in the past, had many great initiatives. One of the things Lisa keeps us grounded on is, how do we go and execute on a vertical, enable a customer, and then go on to the next one? Rather than going too broad and not having enough focus and enough resources to execute.
VB: We’re probably a little ways away from design announcements? You showed some people with products, but they’re not necessarily —
Koduri: On the cloud hyperscale on the compute side, you saw some of the — our current partners have announced their stuff. It’s in their cycle that we announce. We have a lot of good engagement and momentum going on. We’ll see some interesting announcements in 2017. But yes, we’re just going public with our initiative. We’ve been doing a lot of background work and you’ll see the momentum in 2017.
Su: You should expect that we’ll have customer design announcements in 2017.
VB: Do you see the amount of investment that’s gone into both sides, graphics and Zen, as basically the same? Are these equally ambitious projects as far as manpower?
Su: You’re asking me which child I love better, and I’m sitting next to one of them. [laughs] For sure, we’re committed to high performance road maps on both the CPU and the GPU side. I will say, incremental investment in graphics is higher right now. You look at the explosion in the graphics market. It almost has a rebirth going on, not only in terms of consumer but also professional, and now in the data center as well. The incremental investments in graphics around the software and the platform that Raja was talking about are critical for us.
VB: Was it the same years-long effort to get to this point for you guys? Was it coordinated, or did they just happen to appear at the same time, the CPU and GPU?
Koduri: To some extent the same core technology, processor technology, drives both CPU roadmap and GPU roadmap. It’s not coincidental in that sense. A new CPU architecture and a new GPU architecture that leverages the CPU node come together in close proximity. I’d say that’s more a factor than anything else. When there’s a new technology node, all of us are excited, CPU guys and GPU guys. We all try to get something new.
Su: It’s also fair to say that the data center focus has become a clear priority for us over the last couple of years. Having both the CPU and the GPU road map optimized for compute was very important.
Koduri: That’s a huge thing, like I said on stage, that we’re finding with customers, the amount of excitement between the server CPUs and our capabilities there paired up with GPUs. Our customers are telling us, more than us telling them, what the value is there. It’s rare to see that level of excitement from our customers.
VB: Qualcomm had their announcement this week as well, with Snapdragon desktop and laptop SKUs at some point. It’s another interesting avenue of competition.
Koduri: Right. Hardware is hot again.