Our AWS Re:Invent Conference lookahead with AMD’s Brad Smith, Live from Las Vegas

Transcript

Hilary: Welcome to Vegas, Rahul.

Rahul: Welcome to re:Invent, Hilary. This is so awesome.

Hilary: Thank you. I feel like I’m living in the metaverse. This place is crazy.

Rahul: It’s Moore’s Law of crazy.

Hilary: There it is, Vegas. You’ve got your slogan for the week. Let’s do this.

This is AWS Insiders, an original podcast by CloudFix about the services, patterns, and future of cloud computing at AWS. CloudFix is a tool that finds and implements 100% safe, AWS-recommended cost savings, that’s fixes, not just analytics. I’m Hilary Doyle, joined, as always, and now in person, by Rahul Subramaniam. We are coming to you live and on-prem from Las Vegas at re:Invent, sitting up here, outside on the top floor of the Nobu Hotel, overlooking the Vegas strip. It is incredible. That’s about the view that you want.

Rahul: It’s one of the most spectacular locations.

Hilary: Yeah. We’ve heard word that Adele investigated this suite for her residency, passed on it so that we could have it. It was very kind of her. She’s a very generous woman. This is the first of four very special episodes. We’re kicking things off with a conference look ahead with our special guest, Brad Smith from AMD. You’re going to be hearing a lot about AMD over the next couple of days, I suspect. So first, let’s get you caught up.

Rahul: Hilary, thousands of AWS customers, partners, and employees are rolling into Las Vegas. I think the last time I heard it’s going to be about 50 to 60,000 of them.

Hilary: Wow.

Rahul: The thing that I love about re:Invent is the fact that you get to meet people in person and connect with them and build these relationships that you can take back to wherever you come from. We’re looking forward to four days of discussions, announcements, and, of course, the launch, starting with tonight’s first keynote from Peter DeSantis.

Hilary: Yes, VP of global infrastructure, or as the kids call him, the infrastructure czar, although I prefer the Wizard of AWS, pronounced OZ. See what I did there? In the coming days, we’ll cover keynotes from Adam Selipsky, Swami Sivasubramanian, and of course, the legend, the myth, the man in the t-shirt, Werner Vogels.

Rahul: Don’t forget the music, Hilary.

Hilary: I could never forget the music, but Rahul, let’s start here. You have attended almost every single re:Invent. Last year was a 10-year anniversary, so there was a lot of looking back and taking stock, a post-lockdown re-emergence. What is the tone going into this year?

Rahul: I think this year is going to be very interesting and exciting because it’s going to be all about looking forward and for the first time probably, giving guidance on how customers should transform their businesses in the cloud.

Hilary: Well, it’s December. Holidays are coming up. We’re talking a lot about wishlists. If you could have one thing, your dream announcement this year at AWS, what would it be?

Rahul: Okay, I’m going to go all out and crazy on this.

Hilary: Oh, my God.

Rahul: But if there’s one thing that I would like AWS to do is be opinionated and just tell customers exactly what they should do. If they just did that, customers would just love AWS.

Hilary: Listen, it’s just you and me here. Let’s get right to the secret stuff. What inside information have you been able to glean ahead of the keynotes?

Rahul: So AWS is incredibly tight-lipped about literally everything.

Hilary: Yeah, but not with you. I mean, they’re inviting you into their pizza parties.

Rahul: I would say I expect that there’s going to be a lot of announcements around the FinOps stuff, around cost reporting, a lot of simplification happening around that because that’s something that customers have been asking for a whole lot. I think the theme around serverless that picked up steam last year is going to continue. There are going to be a lot more services around serverless. You’re probably going to see a lot more around the new Instance Types that they’re launching, particularly in the machine learning area. But you might see a few more roll-outs of Graviton across the different services. But yeah, overall, I think the theme and the focus is going to be cost, and getting efficiency. So, looking forward to that.

Hilary: Well, it’s interesting. I mean, Vegas is a place that’s completely divorced from reality in so many ways, but we are not. So you’ve mentioned the economic picture playing in the background of this year’s conference. We have heard a lot about massive cuts at Alexa, rising costs from cloud providers as we head into the winter. So this isn’t a rhetorical question for you. Why should we be bullish on AWS?

Rahul: I think what most people don’t get is that cloud computing can be orders of magnitude cheaper than your traditional compute that you have on-premise. It’s just you have to figure out how to leverage it the right way.

Hilary: Sure.

Rahul: The value isn’t in a simple lift-and-shift where you literally just run your servers as is, one-to-one from how you are running them in the data center. The value is all in the higher-order services and the faster you can get to those, the faster you can leverage those, the faster you can start getting more efficient in your cloud cost usage. Wow!

Hilary: I was just hit in the head by our…

Rahul: For our listeners, we just had our billboard, our little board-

Hilary: With our faces on it.

Rahul: With our faces on it.

Hilary: I was hit in the head by my head. That is a perfectly meta Vegas experience. I am going downstairs immediately to bet on red. Do people bet on red here? I don’t know.

Rahul: I have no idea.

Hilary: What are three top predictions for you in terms of announcements?

Rahul: Oh, that’s hard. I think the first one is going to be around simplifying reporting. It’s been a big theme and ask from AWS customers for over a year now, and AWS has acknowledged that. So I think the way they’re going to talk about governance and reporting of all of your cloud resources and your cloud costs, that’s going to be front and center.

The second aspect of it is going to be how to more efficiently make decisions around your AWS Instance Types. As of the last count, they had nearly 500 of them. I think it was 485 or something when I last counted three or four weeks ago, and they announced a few more. So it’s probably going to hit about 500 and you’ve got a problem of choice at this point. So you’ve got to make it easier for customers to pick one, make sure that’s the most efficient one that they’re going to be with, and start leveraging those new Instance Types.

Hilary: I like the sound of that theme and we’ll be keeping our audience up to date as the announcements roll out all week long. We have a tremendous lineup of special guests who are all ready to break down the conversations and the keynotes as they’re happening. Speaking of which, let’s introduce our first guest of the week.

Rahul: Absolutely.

Hilary: Okay. Brad Smith is Corporate VP at AMD. AMD technology powers some of the largest Amazon EC2 customers and internal properties. They’ve recently announced general availability of the 4th Gen AMD EPYC Genoa Processors. Those are for critical workloads across cloud, enterprise, and high-performance computing. Brad Smith, welcome to our podcast. We are so happy to have you. You are our first in-person guest. What an auspicious day.

Brad Smith: Yeah, this is great.

Rahul: And what a venue, right? I mean, we’ve got an amazing outdoor venue over here, so if you hear helicopters and aircrafts flying by, well, that’s life.

Hilary: Watch your head.

Brad Smith: Oh, you’ve got to love Vegas. We were here last year and it was the first time since COVID, so it was a little more subdued. This year, it’s fantastic. You can see everyone flooding in. This place is going to be popping. This is great.

Hilary: Well, we’re going to get right to it because we’ve got a lot to cover with you and we know that your time is limited. AMD has distinguished itself with products across aerospace, entertainment, gaming, automotive, the list goes on. There’s a huge amount of buzz right now around the Genoa Processor. Would you give us just the high-level on Genoa for anyone who’s missed the memo?

Brad Smith: Sure. In simple form, it’s the next volley in our EPYC product line for the data center, and it’s built on 4th Gen core and it’s a beast. It’s 96 cores built on PCIe Gen 5 and it’s got 12 channels of DDR5. AWS recently announced the support for the product and we’re looking forward to future releases based on that. We’re very excited about that product. It’s got so many various things you can do with it. So right now, the problem is, what are we going to do with it? But it’s the most performant x86, the most power-efficient x86 offering on the planet.

Rahul: I’ve got to ask this, Brad. I’ve always been curious about where all the server technology is headed and I’m very curious about how AMD sees things. I see three vectors. One is, you keep increasing the clock cycles on your processors and somehow I feel like we are hitting an upper bound. The five gigahertz is about as far as you can push it without melting the chips on the boards. The second one is the number of cores that you put on a single chip, and then the third one is how many sockets you put on a motherboard. So how do you see AMD seeing the future when it comes to those three dimensions?

Brad Smith: We love the position we’re in right now in the market. We are heavily focused, obviously, in performance, but we’re also focused in power efficiency. It’s a big part of our model. We’re trying to ensure that everything that is released, it’s better for the planet than it was previously. So we’re a leader in that space.

Think about cloud competing with AWS. They like a common platform that they can configure in various ways. So you talked about the 500 instances. They certainly don’t have that many products integrated in the environment. So they’re very, very good at taking, not a Swiss Army knife to things, but somewhat of that approach, whether it’s memory bound or compute bound or it’s network, a bandwidth heavier, or EBS bandwidth heavy. So it just depends on the workload. I think over time, you’ll see a lot more innovation in semiconducting geared towards the type of workloads that are being run, and I think historically, it was more just about power, power, power, and efficiency, efficiency.

As the software stack starts to compress downward into hardware, we’ll continue to see that cycle. I think you’ll start to see a lot more innovation. This particular workload does things that nothing else can, right on that particular product, and whether you’re talking about microservices and containerized workloads, you’re talking about pure classic web server, web hosting, you’re talking about database. There’s so many ways that a company like AMD can go now.

Hilary: You’ve alluded to this already, but could you tell us a bit more how you expect hardware to influence software in the coming year?

Brad Smith: I honestly never even thought twice about a CPU until I came to AMD and it opened my eyes to understand what can be done. So even simple tweaks with your compiler can generate an additional 10 to 15% on your performance. You guys talked about cost optimization earlier. AMD already has the same performant processor on our Milan generation at 10% less cost. So you’re already, for no uplift at all, saving 10% of your compute, which is a big deal.

Now you start to layer in some tweaks to your software along the same lines, you can get another 15 to 20% just on that particular workload. So now you’re talking a 25 to 30% cost benefit. Take that over a three-year period knowing that half of every spend is compute, at least. You’re talking about tremendous savings, and it’s just one of those things that most companies don’t understand. It’s one of the things I love about CloudFix, is your message is so on-point right now and it’s also very simple. It’s very simple. Your tool is not simple when it comes to what it has to do, but the way it presents itself is very simple.

Hilary: We’ve got to wrap this up and I’m so sad about that because we’d love to sit here and chat with you all day. Quickly, what are the three talks you’re looking most forward to?

Brad Smith: Oh, definitely Peter DeSantis. I love his talks and he gives a lot of insight into Graviton, which is great. I love to see what’s going on, and AMD loves competition. I think we’ve solidified ourselves as we are the x86 platform of choice right now, and they’re pushing very much for Arm and they’re very complimentary together.

So it’s an interesting relationship as it starts to move forward. Peter’s fantastic. Of course, Adam is great, and Werner is a classic. I’ve been watching him on stage for so long. It’s funny to watch how it evolves, how the message evolves. The rest of it, I love to go into the smaller venues and the topics that start to interact with everybody and just mix about. So I think those are probably the three I hit, and outside of that, the rest will just be conversations.

Hilary: Okay. We’ll see you on the floor. Thank you so much for joining us, Brad, a real pleasure to meet you.

Brad Smith: My pleasure. Thanks for having me.

Rahul: Thanks for being here, Brad.

Brad Smith: Thank you.

Rahul: An absolute pleasure.

Hilary: Rahul, AMD has had an incredible trajectory over the last couple of years and what they’ve done with the Genoa chip is a pretty apt metaphor for the benefits of moving to the cloud. Would you give us the background on the UFC-style rivalry between AMD and Intel? I don’t think it’s actually a rivalry, but we’re in Vegas, so I’m bringing the drama.

Rahul: Oh, it’s totally a rivalry.

Hilary: Okay.

Rahul: So here it goes. Till about 2012, 2013, AMD was always playing catch up to Intel. Intel always had the better processors and AMD was trying to catch up, and it was all about the fabrication units that they had. Come 2013, 2014, I think, is when they started making decisions to take all of their fabrication and give it to TSMC and leverage all of the amazing stuff that TSMC was doing with new chip manufacturing technology. Intel, on the other hand, decided to have their own fabrication units, very similar to the cloud story.

Hilary: Just go on. And Intel has been late on all of the chip deliveries, right?

Rahul: Correct. Just keeping up with all the new technology that’s coming up on fabrication and trying to build their own, it just keeps them behind on everything. AMD has been incredibly smart in just focusing on the designs of these new chips and getting TSMC to manufacture it. So they are winning big. They are really, really kicking Intel’s arse on this.

Hilary: All right. We like winning big in Vegas. Listen, it’s morning here and you have already been up for about 14 hours. So one final thought before you fall asleep at the table.

Rahul: I’d say, look out for all the amazing announcements around the new Instance Types. If not at re:Invent, shortly after. I can’t wait to see what’s going to happen with the new general processors and new insights that come out of it. So yeah, a lot of exciting stuff coming around the instances.

Hilary: Okay, that’s it for day one. Much more to come. You can join us tomorrow for the reaction to Adam Selipsky’s keynote.

Rahul: Send us your thoughts, queries, reactions, betting requests…

Hilary: …you name it, to podcast@CloudFix.com.

Rahul: Bye-bye for now. We’ll see you soon.

Hilary: We’ll catch you later.