>>from around the globe. It's the Cube with digital coverage of AWS reinvent 2020 sponsored by Intel, AWS and our community partners. Everyone welcome back to the cubes. Coverage of AWS Reinvent Amazon Web services Annual conference theme. Cuba's normally there in person. This year we can't be. It's a virtual event. This is the Cube virtual. I'm your host for the Cube. John Ferrier Tresh Damn Kroger, VP of G M and G m of the high performance computing team at Intel is here in the Cube until a big part of the cube every year. Trish, thank you for coming on Were remote. We can't be in person. Um, good to see you. >>Good to see you. >>I'm really impressed with Reinvent Has grown from kind of small show eight years ago to now kind of a bellwether. And and every year it's the same story. More scale, more performance, lower prices. This is kind of the intel cadence that we've seen of Intel over the years. But high performance computing, which >>has been >>around for a while, has gotten much more mainstream thinking because it's applying now to scale. So I want to get your thoughts and and just set the table real quick. What is high performance computing mean these days from Intel? And has that relate to what people are experiencing >>e high performance computing? Um, yes, it's been traditionally known as something that's, you know, in the in the labs and the government, you know, not used widely. But high performance computing is truly just changing the world is what you can dio Cove. It is a great example of where they've taken high performance computing to speed up the discovery of drugs and vaccines for or cova 19. They use it every day. You know, whether it's making Pampers or Clorox boxes. So they are those bottles so that they, when you drop them, they don't break, um, to designing airplanes and designing, um, Caterpillar tractors. So it is pervasive throughout. And, um, sometimes people don't realize that high performance computing infrastructure is kind of that basics that you use whenever you need to do something with dense compute. >>So what some examples of workloads can you just share? I mean, obviously Xeon processor. We've covered that many times, but I mean from a workload standpoint, what kind of workloads are high performance computing kind of related or unable or ideal for that's out there, >>right? Z on scalable processors are the foundation for high performance computing. If you look at what most people run high performance computing on its see on, and I think that it's so broad. So if you look at seismic processing or molecular dynamics for the drug discovery type work or if you think about, um, open foam for fluid dynamics or, um, you know, different financial trade service, you know, frequency, fats, frequency trading or low. I can't even think of that word. But anyway, trading is very common using high performance computing. I mean, it's just used pervasively throughout. >>Yeah, and you're seeing you're seeing the cloud of clarification of that. I want to get your thoughts. The next question is, you know it's not just Intel hardware. You mentioned Zeon, but HBC in AWS were here. It reinvent. Can you share how that plays out? What's your what's your What's your take on that? Because it's not just hard work and you just take them into explain relationship, >>right? So we definitely have seen the growth of high performance computing in the cloud over the last couple of years. We've talked about this for, you know, probably a decade, and we've definitely seen that shift. And with AWS, we have this wonderful partnership where Intel is not only bringing the hardware like you say, the Z on scalable processors, but we're also having accelerators and then on that whole software ecosystem where we work closely with our I s V and O S v partners. And when we bring, um, not only compilers but also analyzers in our full to tool suite so people can move between an on Prem situation Thio Public cloud like aws. Um, seamlessly. >>So talk about the developer impact. As I say, it's that learning show reinvent. There's a lot of developers here. I'll see mainstream you're seeing, you know, obviously the born in the cloud. But now you're seeing large scale enterprises and big businesses. You mentioned financial services from high frequency trading to oil and gas. Every vertical has a need for cloud and and what, you should be traditionally on premises compute. So you have. You're kind of connecting those dots here with AWS. Um, what is some of the developer angle here? Because they're in the cloud to they want to develop. How does how does the developer, um, engage with you guys on HPC in Amazon, >>Right? Well, there's there's a couple ways. I mean, so we do work with some of our partners eso that they could help move those workloads to the cloud. So an example is 69 which recently helped a customer successfully port a customized version of the in car models for prediction across scales. So they chose the C 59 18 x large instance type because this is what really deliver the highest performance and the lowest price for compute ratio. Another great example is P. K. I, which is a partner out of the UK, has worked with our customers to implement AI in retail and other segments running on Intel Instances of the EEC too. So I think these air just so you could have people help you migrate your workloads into the cloud. But then also, one of the great things I would like to talk about is, um a ws has come out with the parallel cluster, which is an Intel select solution, which really helps, um, ease that transition from on Prem to cloud. >>That's awesome. Um, let's get into that parallel cluster and you mentioned Intel Select Solution program. There's been some buzz on that. Can you take a minute to explain what that is? I >>mean, the HBC has, AH reputation of being hard, and the whole philosophy between behind the Intel Select solution is to make it easier for our customers to run HBC workloads in the cloud or on Prem and with E Intel Select Solution. It's also about scaling your job across a large number of notes, so we've made it a significant investment into the full stack. So this is from the silicon level all the way up to the application level so that we ensure that your application runs best on Intel and we bring together all the everything that you need into. Basically, it's a reference design. So it's a recipe where we jointly created it with our I, C, P and O S V partners and our open source environment for all the different relevant workloads. And so Amazon Web Services is the first cloud service provider to actually verify a service such as Intel Select Solution and this is this is amazing because this truly means that somebody can say it works today on Prem, and I know it will work exactly the same in AWS Cloud. >>That's huge. And I wanna just call that out because I think it's worth noting. You guys just don't throw this around like in the industry like doing these kind of partnerships. Intel's been pretty hard core on the quality, and so having a cloud service provider kind of go through the thing, it's really notable you mentioned parallel cluster um, deal. What is Can you just tie that together? Because if I get this right, the Intel, uh, select solution with the cloud service provider Amazon is a reference designed for how to go on premise or edge or revenue. It is to cloud in and out of cloud. How does this parallel cluster project fit into all this? Can you just unpack that a little bit? >>Right. So the parallel cluster basically, um, it's a parallel cluster until select solution. And there's three instances that we're featuring with the Intel Xeon Scalable processor, which gives you a variety of compute characteristics. So the select solution gives you the compute, the storage, the memory the networking that you need. You know, it says the specifications for what you need to run a non optimal way. And then a WS has allowed us to take some of the C five or some of the instances, and we are on. Three different instances were on the C five, in instance. But that's for your compute optimize work clothes. We're on the in five instance and that's really for a balanced between higher memory per core ratio. And then you have your are five and instance at a W s that's really targeted for that memory intensive workloads. And so all of these are accessible within the single A. W s parallel cholesterol environment on bits at scale. And it's really you're choosing of what you want to take and do. And then on top of that, the they're enabled with the next generation AWS Nitro system, which delivers 100 gigabits of networking for the HBC workloads. So that is huge for HPC. >>I was gonna get to the Nitro is my one of my top questions. Thanks >>for >>thanks for clarifying that. You know, I'm old enough to remember the old days when you have the intel inside the PC a shell of, ah box and create all that great productivity value. But with cloud, it's almost like we're seeing that again. You just hit on some key points you have. Yeah, this is HPC is like memory storage. You've got networking a compute. All these things kind of all kind of working together. If I get that right, you just kind of laid that out there. And it's not an intel Has to be intel. Everything. Your intel inside the cloud now and on premise, which is the There is no on premise anymore. It's cloud operations. If I get this right because you're essentially bridging the two worlds with the chips, you bring on premise which could be edge a big edge or small legend in cloud. Is that right? I mean, this is kind of where this is >>going. Yeah, so I mean, what I think about so a lot of them. The usages for HBC in the cloud is burst capacity. Most HBC centers are 100% not 100% because they have to do maintenance, but 95% utilized, so there is no more space. And so when you have a need to do a larger run or you need thio, you know, have something done quickly you burst to the cloud. That's just what you need to do now. I mean, or you want to try out different instances. So you want to see whether maybe that memory intensive workload would work better? Maybe in kind of that are five in instance, and that gives you that opportunity to see and also, you know, maybe what you want to purchase. So truly, we're entering this hybrid cloud bottle where you can't, um the demand for high performance computing is so large that you've got to be able to burst to the cloud. >>I think you guys got it right. I'm really impressed. And I like what I'm seeing. And I think you talked about earlier the top of the interview, government labs and whatnot. I think those are the early adopters because when they need more power and they usually don't have a lot of big budgets, a little max out and then go to the cloud Whether it's, you know, computing, you know what's going on in the ocean and climate change are all these things that they work on that need massive compute and power. That's a a pretext to enterprise. So if you can't connect the dots, you're kind of right in line with what we're seeing. So super impressive. Thanks for sharing that. Final thoughts on this is that performance. So Okay, the next question is, OK, all great. You're looking good off the tee or looking down the road. Clear path to success in the future. How does the performance compare in the cloud versus on premise? >>It could be well, and that's one of the great things about the Intel select solution because we have optimized that reference designed so that you can get the performance you're used to on Prem in the AWS Cloud. And so that is what's so cool honestly, about this opportunity So we can help you know, that small and medium business that doesn't maybe have this resource is or even those industries that do. And they know they're already a reference using that modeling SIM reference design, and they can now just burst to the cloud and it will work. But the performance they expect >>Trish, great to have you on great insight. Thanks for sharing all the great goodness from Intel and the A W s final thoughts on the on the partnership. We're not in person. And by the way, Intel usually has a huge presence. The booth is usually right behind the cube stage, which you guys sponsor. Thank you very much greater. Always partner with you. Great party. You sponsor the replay, which is always great, and it's always great party and great partnership. Good content. We're not there this year. What's the relationship like? And you take a minute to explain your final thoughts on a Amazon Web services and intel. >>Yeah, I know we have, Ah, Long term partnership 14 plus year partnership with AWS. And I mean, I think it's with the your, um taking Intel Select solution. It's going to be even a richer partnership we're gonna have in the future. So I'm thrilled that I have the opportunity to talk about it and really talk about how excited I am to be able Thio bring Mawr HBC into the world. It's all about the democratization of HBC because HBC changes the world >>well. Tricia, congratulations on the select program with AWS and the first cloud service provider really is a nice directional indicator of what's gonna happen. Futures laid out. Of course. Intel's in front. Thank you for coming. I appreciate it. >>Oh, thank you, John. >>Okay, that's the cubes. Virtual coverage Cube. Virtual. We're not in person. Aws reinvent 2020 is virtual. Three weeks were over the next three weeks, we're gonna bring you coverage. Of course. Cube Live in studio in Palo Alto will be covering a lot of the news. Stay with us from or coverage after this short break. Thank you.