AMD’s Lisa Su at Computex, on the record: Threadripper, Huawei, ray tracing and more
Su addresses Threadripper, the Huawei situation, motherboard BIOS issues, and why anyone should care that Intel is hiring so many journalists.
By Mark Hachman
PCWorldMay 27, 2019 9:52 pm PDT
Image: Adam Patrick Murray / IDG
After AMD chief executive Lisa Su wowed a Computex audience with her announcement of a cost-effective Ryzen 9 3900X, we were still left with some questions: Where was Threadripper, for example? In a question-and-answer session following her speech, Su took the time to address this and many other topics, from ray tracing to Huawei and more.
Note: Like all transcripts, there can be crosstalk, coughing, confusingly worded statements, and so on, which we’ve trimmed where needed for clarity. Journalists also tend to jump around from topic to topic. While we’ve tried to transcribe Su’s remarks verbatim, we’ve edited the questions from the journalists. We’ve also used “reporter” to indicate a question asked by any reporter. Finally, even a 30-minute transcript amounts to an enormous block of text, so use our Table of Contents links to jump from highlight to highlight.
Addressing rumors, and Threadripper
Reporter: While I think the 3900X is a great part, there are people on the Internet who expected to see a 16-core part. So Lisa, where’s the 16-core part?
Lisa Su: You don’t think we had enough content? You know how much stuff we took out? Look, we are extremely proud of third-gen Ryzen. When you look at every aspect of it, you know, the core, the system, all that stuff—I think you know that we always push the envelope. So I would say that we’re extremely excited with the lineup we announced today. And when there’s more we’ll tell you about more.
Reporter: There was no mention of Threadripper today. Can you give us any sort of update on that part?
Su: Yeah, you know, it’s very interesting, some of these things that circulate on the Internet. I don’t think we ever said Threadripper was not going to continue. It somehow took on a life of its own on the Internet.
You will see more Threadrippers from us. You will definitely see more. Look, we love the high-end desktop market. I think we’ll see that, both for content creators as well as workstation needs, Threadripper has done well. And so you will see more from us with Threadripper.
Reporter: There’s so much passion in the community. When rumors get started, do they sort of end up out of control? Do you want to say anything to that crowd?
Su: First, what I would like to say is, I read more than any of those guys think. So look, we are very flattered that there are so many people who wonder what we’re doing, whether it’s Ryzen, or Navi, or, you know, what’s the IPC [instructions per clock], what’s the core count? What’s the frequency? What’s the price? We have so much advice that people have been given us. And all I can say is the community’s very, very important to us, very important to each one of us, and what we believe is that we’re giving the community an exceptional set of products. And that’s how we feel about it.
Do benchmarks still matter?
Reporter: Last night, Intel made a big pitch, part of which is that reviewers are using artificial benchmarks, benchmarks that don’t reflect the real world. And they’re starting to try to influence the community to try and move away from that. How do you feel about that?
Su: We also believe that real-world applications are what’s important, no doubt about it. But at some point, you’ve got to compare X to Y. So we will use benchmarks, we do. You might have noticed that we switched from Cinebench R15 to R20. We did that on purpose, because we thought it was a harder test, frankly, than R15. Maybe David or Robert can comment more about it. When we look at gaming performance, we do our very best to benchmark. All of our stuff is apples to apples, and we’ll continue to do that.
Benchmarks are important, they give you a view of competitiveness. But at the end of the day, it’s the user at home. And what we believe is we’ve given the user a lot of choice, depending on where your price points are, what your performance requirements are, whether you want to build up, you know, you want to use a water cooler or an air cooler. I think we’ve given you a lot of choice in the processor capabilities.
David McAfee, senior director of product management: I’ll add just a little bit to that on what you saw. The Ryzen processors today, Cinebench R20 is based on the same engine that’s used in Maxon Cinema 4D, which is a real-world application that tasks processor performance. And then our second demonstration was built around Blender, which again, is a real-world application that people use every day. We’ve tried very, very hard to make sure that as we talked about the performance of our products, it’s based on real experiences that people are going to get and not synthetic workloads that are representative of what’s missing.
Robert Hallock, AMD’s senior technical marketing manager for CPUs: Yeah, this topic is near and dear to my heart. My day job is picking the benchmarks that are used to promote Ryzen. The last thing that I would ever want to do in my capacity at AMD is put forth a set of benchmarks that misleads the public in any way, that makes them feel like we’re out of line with what they want. And so we’re constantly looking at the use of benchmarks in the reviewer community, that users use, that a lot of hardware companies use—we try to find that right balance. So we’re authentic and honest with what the product is. If you do anything else it’s just an injustice to both the community and the product itself.
Epyc issues, and 5nm
Reporter: I have a question about your Epyc data center server processor. Do you have a market share target that you could elaborate upon?
Su: Look, we are always looking to increase our market share. That’s why we put out great products. As it relates to our market share targets—for server, what we’ve said is, we believe that we can achieve double-digit market share, four to six quarters from sort of the end of 2018. So that was sort of the timing of that.
If you talk about graphics, you know, I’ll let these guys talk about graphics. The truth is AMD graphics has been very strong in the past, and it’s very strong today. And with the RDNA architecture, we think it will be even stronger going forward. So we’re very excited about opportunities to gain share in graphics.
And then in the PC market. You know, when you look historically at AMD in the PC market, we have been about high teens, low double-digits market share depending on time. We’ve gained market share the last six quarters—is that right?— so six quarters, we’ve gained PC market share on the strength of Ryzen. We think third-gen Ryzen is going to be very helpful to continue that.
Reporter: TSMC is a very important ecosystem partner right now. Can you share with us how much is going to TSMC, and when you plan to move to 5 nanometer?
Su: So we’ve done many, many, 7nm products, I don’t think we have said exactly how many. But think about it as server, PC, graphics, and as well as our custom products. So it’s a broad product portfolio. And, you know, we’re going to be aggressive with leading-edge technology. So I’m not saying which five nanometer and when, but you will see us aggressively taking the initiative.
More details on ray tracing?
Reporter: With ray tracing apparently being a big part of future consoles, is it going to be part of [the new graphic architecture], RDNA?
Su: So we have a lot more RDNA content that David and Scott will be presenting. I think some of you may be coming to our tech day, in a few weeks at E3. We’re going to talk about all of that.
Scott Herkelman, AMD Radeon general manager: We’ll give out a lot more detail on that, all the new features and technologies.
Su: We only had an hour today.
Reporter: It seems like if ray tracing was a thing you were definitely working on, we’d have heard it on stage today?
Su: We’re definitely working on ray tracing. That’s true. But like I said, we’ll give you more of the roadmap at E3.
Reporter: Given that Sony has already announced ray tracing as part of the PlayStation 5, can you tell us if that’s a Sony optimization, or part of RDNA?
Su: So we certainly have done very specific optimizations for Sony. They are a deep customer for us on semi-custom products. There are optimizations there. However, we view ray tracing as a very important element across the portfolio. So we’ll have ray tracing a number of other places… Look at that, you got me to say more about ray tracing!
David Wang, senior vice president of engineering at AMD: We started our RDNA development before the Sony engagement. RDNA is a revolutionary architecture; it’s also very flexible. So it can be optimized [inaudible].
Reporter: So it’s like an FPGA.
Su: I wouldn’t exactly say that.
Reporter: Followup question: Sony, Microsoft, and Google’s Stadia have announced cloud gaming initiatives. Since AMD powers all the “on-premise” game consoles, how do you see cloud gaming stacking up against local consoles?
Su: Well, look, I think we believe gaming will be all form factors. So whether it’s PCs, or consoles, or cloud, all of them require great graphics capability. As well as a number of other things. We’re very proud to be partnering with Google on their Stadia cloud streaming platform; there are a number of other cloud efforts that we are very involved in. And, you know, from my standpoint, I think they’re all going to coexist. So it’s not like one is going to take over. Their business models, all that stuff. But I think cloud gaming is going to be important. And you know, we’re going to continue to invest in that from a technology standpoint.
Reporter: Is that an Epyc play?
Su: It’s Epyc and Radeon.
Reporter: I do feel like a lot of what we saw in the PlayStation 5 hinted at a lot of things you guys were working on. A big thing for them was no loading of levels. Was there anything in particular that could allow you to bring that to the PC market? PCI Express 4?
Su: PCIe 4 definitely helps. Things that Sony are doing, they’ve been very specific on their proprietary technologies.
Why hasn’t AMD jumped into AI? That and more on the next page.
Why hasn’t AMD announced anything in AI?
Reporter: We’ve seen a lot from other companies in the industry, that one of the buzzwords over the next few years is AI machine learning. Can you talk to what AMD is doing in this area, especially when a lot of your competitors are actively engaged here?
Su: Yeah, so we really believe in a heterogeneous architecture. So if you look at our consumer CPUs, or sort of how we think about, you know, sort of the system environment, you know, the world of CPUs plus GPUs, plus accelerators are important. We are very actively working on some of these machine-learning accelerators as well being integrated into our silicon. And we’ll talk more about that as we get closer.
Reporter (who arrived late): David, what are your thoughts about ray tracing in mainstream graphics?
Wang: We’ll talk more about it in June.
Su: Ten days.
McAfee: We’ll give you a good full update on the tech day.
Su: What you should expect, though, is ray tracing, it’s important technology, you will see it across our portfolio, and particularly in working with the ecosystem, we’ll ensure that there there will be strong ecosystem support.
Reporter: A slightly different question: What are your thoughts on your competitors’ hiring up all the tech journalists? It means fewer people talking about AMD.
Su: You guys seem to talk about us a lot, so I don’t feel worried about that. Look, I am extremely proud of the team that we have at AMD. And if you look across our engineering team, you know, on the CPU side, the GPU side, the business environment, I think we have a tremendous team that’s always going to push the envelope. I like tech reporters reporting on technology. So thank you, and you guys for doing that. And, you know, I feel very, very good about the amount of attention that AMD gets. Sometimes, I think, maybe a little too much attention.
No, actually, that’s not true. There’s no such thing as too much attention. There’s probably a little bit of ensuring that the rumor mill doesn’t get out of control.
Reporter: Now that you’ve got more cores in mainstream Ryzen, can it be argued at all that it’s stepping on Threadripper’s toes?
Su: I was asked that question earlier. I think Threadripper is still important. So you will see future generations of Threadrippers from us. Now, obviously, if mainstream is moving up, then Threadripper is going to have to move up-up. That’s what we’re working on.
Reporter: It feels like Intel has hit a wall with desktop clients, and that they’re emphasizing platform technologies like Optane and Thunderbolt as much as anything else. Is that going to be a problem for AMD?
Su: Well, first of all, we haven’t hit a wall on desktop architectures. And we’ve been really sort of focused on bringing more compute performance into whether you’re talking about desktop, or you’re talking about the server ecosystem or any of these ecosystems. I think open standards are really important. And we work across the industry, with the memory vendors, and all of the other guys, and we’re going to continue to do that. So I don’t see it as a significant disadvantage, I see it as an opportunity. And that’s why we’re always bringing the ecosystem together.
If you look at the number of people who are supporting PCI Gen4, we didn’t expect to be first. The fact that we are first and all of these ecosystem partners are extraordinarily aggressive getting their technology qualified and available, to sit with our ecosystem, I think gives you just a little bit of a view of the [inaudible] ecosystem.
Reporter: With PCI Express 4, there was a pretty big lag before that was introduced in the market. PCI Express 5 is just around the corner. Do you see a faster uptake of that, and do you think PCI Express 4 will be short-lived?
Su: Well, I think it remains to be seen, things always seem like they’re going to be faster. And things always take a little bit longer than you expect.
Reporter: As a follow up, there have been reports that some motherboards won’t support the new Ryzen, specifically the third-gen processors. Is that accurate? The A-series boards in particular.
McAfee: Right. So I think that as we look across the ecosystem of motherboards that exists today, we certainly make available BIOS updates to our ecosystem partners to include that on different levels and on the boards that they have in their portfolio. But I do not expect that every motherboard will be updated for 3000-series processors [inaudible]. That really will be a portfolio decision on their standpoint as well as to where they apply those updates. And when they choose not to apply those updates.
Hallock: I do think there is a bigger story here, that no one in the history of x86 has created an upgradable socket quite like [AMD’s socket] AM4. In a time where our competitor is breaking socket compatibility yearly—basically, you have three consecutive generations, all dropped into the same socket. And that socket was started at four cores, years ago, and is now 12 cores 24 threads PCI Gen 4, [inaudible]. And that to me is the bigger story than a motherboard here and there that doesn’t get a BIOS update.
Reporter: With PCI Express 4.0 backwards compatibility, who is validating that and the transfer rates?
McAfee: So certainly if if they’re going to declare their motherboards PCI Gen4 ready, that’s a certification that happens through the PCI SIG, not through AMD.
I think that you know, what we have seen as we launch our new product, our X570 platform, that’s the platform that we carry the PCIe Gen4 readiness certification that goes along with it. Beyond that, other motherboards may or may not be compatible with Gen4. So it really depends on how those old boards were designed, what their capabilities are at a platform level. And as we go out the gate, we do not expect the older motherboards to have compatibility for Gen4.
Reporter: Can we say anything about Renoir [a rumored APU combining Zen 2 technology and Vega GPU cores]? There have been reports that the project is dead.
Su: That is not true. It is doing well.
Drew Prairie, AMD spokesman: We might want to start with, we don’t know what Renoir is.
Reporter: Future Ryzen mobile.
Huawei and the PC market
Reporter: Given that the entire PC industry is worried about the future, are you? Does that weigh on your outlook?
Su: From a product standpoint, it does not change what we do at all. From a product standpoint, our roadmaps are laid out for the next three to five years. And we’re continuing to be very aggressive on that roadmap.
From a business standpoint, of course, we pay attention to some of the global issues that are out there. And, you know, all business leaders would like it to be resolved as soon as possible. But from our product roadmap standpoint, it does not have a significant impact.
Reporter: Can you comment on your relationship with Huawei?
Su: Huawei is a customer of ours, they’ve done some very nice PCs, with our first-generation Ryzen as well as our second-generation Ryzen. You know, obviously, we’re a U.S. company. So we’re complying with the current U.S. regulations. And, as I said yesterday, we would like things to be resolved as soon as possible.
Reporter: So does this mean that you are suspending sales?
Su: We are abiding by the U.S. regulations.
Is AMD adding more design teams?
Reporter: I’ve been told by Mark Papermaster that there’s one CPU architecture group and two implementation groups. Given your recent successes, is that expected to change?
Su: The CPU group is definitely getting larger. By the way, so is the GPU group. David has a very large team. We are on a very clear roadmap. To Zen 2, which we’re talking about today, Zen 3 is deep, deep into development, Zen 4 is also in development and similarly we’re talking about the first generation of RDNA today, we have several generations in parallel that are working.
So I think the one thing that I would like to say is our roadmap has not changed. When Mark and I and the rest of the team started the roadmap, it was with the idea that we needed multiple generations of continuous improvement. So for Zens and plus, and twos, and three, that was all part of the plan. And as we go forward, we’re going to continue to be very, very aggressive on both the CPU and GPU engines.
Reporter: So I guess my question was whether AMD has the ability to bring two separate CPU architectures into the market.
Su: You know, it’s not clear that we would want to do that. What we have learned is by focusing our resources on Zen—of course, you understand that the Zen that we brought in server is a bit different from the Zen that we brought in PC form factor. Certainly the frequencies, the powers, all those things are a little bit different. But we get incredible learning by using the same technology across multiple markets. So I don’t think we are likely to change that in the near future.
Thoughts on THATIC
Reporter: Going back to the U.S. regulations question, how is your joint venture with THATIC affected?
Su: So THATIC was formed several years ago. We did the initial technology transfer, at that point of time. We are continuing the joint venture, and most of the work was on the joint venture side, not the AMD side.
Reporter: Will that continue moving forward into future generations such as Zen 2 or Zen 3?
Su: We are not discussing any additional technology transfers.
Reporter: You aren’t discussing with them, or you aren’t discussing with us?
Su: Let me be clear. The THATIC joint venture was a single generation of technology devices. There are no additional technology licenses.
Reporter: Did they license a core, or…?
Su: A single implementation. I don’t think we ever said what they licensed. We said they licensed an x86 CPU implementation.
Does the future of the PC include AMD?
Reporter: On a broader level, I’m wondering, how are you looking at the changing form factors of PCs? Is that something you guys are thinking about?
Su: We’re doing a lot with the PC OEMs. As part of the newer form factors are being done with second-generation Ryzen Mobile, I think you’ll see more as we go forward. So yeah, look, I think PC, form-factor innovation is pretty important. And we’re very involved with both Microsoft and the OEMs.
Reporter: How important is the halo spot in the GPU market for AMD?
Su: You should ask David that.
Wang: Very important. We should be able to compete very well in the higher-end space.
Reporter: How do you feel about 5G PCs? Do you think that there will be 5G PCs?
Su: Yeah, look, I think that there will be 5G PCs. Certainly.
I think 5G is still early. Also, it’s still the infrastructure timing at this point, not quite [there] on the consumer side.
Reporter: Intel loves doing these projects, such as Project Athena, which are designed to bring an ecosystem together but also involve selling more Intel hardware…
Su: I think, look, what you hopefully have seen is the progression of form factors from first-gen Ryzen to second-gen Ryzen to beyond second-gen Ryzen. And I will say that, for a long time, OEMs didn’t necessarily put AMD processors in the best form factors. And we’ve been working very closely with the OEMs to introduce many more. That’s why you saw, sort of the Microsoft modern device category as being something that we focused a lot on.
We’re focusing a lot on the user experience, and what that brings, and I think you will be very pleasantly surprised with new form factors that will come out with AMD in the coming months.
Reporter: Given that AMD is exclusively sourcing 7nm from TSMC, with no redundancy from GlobalFoundries, are you worried about supply at all?
Su: We have a great relationship with TSMC. And I would say that our rev for 7nm has been one of the smoothest revs that we’ve had across a number of different parts. So yes, I believe that we’re going to have the capacity that we need.
Prairie: I think that’s about all the time we have. Thanks, everyone.