 Welcome back, everyone, live here at Supercomputing22 in Dallas, Texas. I'm John Furrier, host of theCUBE, here with Paul Gillan, editor of SiliconANGLE. Getting all the stories, bringing it to you live, SupercomputerTV is theCUBE right now, and bringing all the action. And Kurt Brezniker, chief architect of ULT Packard Labs, with HPE CUBE alumnus here, to talk about supercomputing, road to quantum. Kurt, great to see you. Thanks for coming on. Thanks for having me, guys. It's great to be here. So Paul and I were talking, and we've been covering computing as we get into the large-scale cloud, on-premises. Compute has been one of those things that just never stops. No one ever, I never heard someone say, I want to run my application or workload on slower hardware or processors or horsepower. Computing continues to go, but we're at a step function. It feels like we're at a level where we're going to unleash new creativity, new use cases. You've been kind of working on this for many, many years at HPE ULT Packard Labs. I remember the machine and all the predecessor R&D. Where are we right now, from your standpoint, HPE standpoint? Where are you in the computing? It's as a service. Everything's changing. What's your view? So I think, you know, you capture so well. You think of the capabilities that you create. You create these systems and you engineer these amazing products and then you think, whew, doesn't get any better than that. And then you remind yourself as engineer, but wait, actually it has to, right? It has to because we need to continuously provide that next generation of scientists and engineer and artist and leader with the tools that can do more and do more, frankly, with less because while we want to run the program slower, we sure do want to run them for less energy and figuring out how we accomplish all of those things I think is really where it's going to be fascinating. And it's also, you know, we think about that. We think about that now exascale data center, billion, billion operations per second, the new science, arts and engineering that we'll create. And yet it's also what's beyond. What's beyond that data center? How do we hook it up to those fantastic scientific instruments that are capable to generate so much information? We need to understand how we couple all of those things together. So I agree, we are at an amazing opportunity to raise the aspirations of the next generation. At the same time, we have to think about what's coming next in terms of the technology. Is the silicon the only answer for us to continue to advance? You know, one of the big conversations is like refactoring, replatforming. We have a booth behind us that's doing energy. We can build it in data centers for compute. There's all kinds of new things. Is there anything in the paradigm of computing and on the road to quantum, which I know you're involved. I saw you on LinkedIn, you have an open rec for that. What paradigm elements are changing that weren't in play a few years ago that you're looking at right now as you look at the 20 mile stair into quantum? So I think for us, it's fascinating because we've had a tailwind at our backs my whole career, 33 years at HP. And what I could count on was transistors at first they got cheaper, faster, and they used less energy. And then that slowed down a little bit. Now they're still cheaper and faster as we look in that and that Moore's Law continues to flatten out of it. There has to be something better to do than yet another copy of the prior design opening up that diversity of approach. And whether it is the amazing wafer scale accelerators we see, these applications specific silicon, and then broadening out even farther next to the silicon. Here's the analog computational accelerator. Here is now the emergence of a potential quantum accelerator. So seeing that diversity of approaches, but what we have to happen is we need to harness all of those efficiencies. And yet we still have to realize that there are human beings that need to create the application. So how do we bridge? How do we accommodate the physics of new kinds of accelerator? How do we imagine the cyber physical connection to the rest of the supercomputer? And then finally, how do we bridge that productivity gap? Especially not for people who like me who've been around for a long time. We want to think about that next generation because they're the ones indeed that solve the problems and write the code that will do it. You mentioned what exists beyond silicon. In fact, are you looking at different kinds of materials that computers in the future will be built upon? Oh, absolutely. You think of when we look at the quantum modalities and whether it is a trapped ion or a superconducting piece of silicon or it is a neutral ion. There's just no, there's about a half a dozen of these novel systems because really what we're doing when we're using a quantum mechanical computer, we're creating a tiny universe. We're putting a little bit of material in there and we're manipulating at the subatomic level, harnessing the power of quantum physics. That's an incredible challenge and it will take novel materials, novel capabilities that we aren't just used to seeing not many people have a helium supplier in their data center today, but some of them might tomorrow and understanding, again, how do we incorporate industrialized and then scale all of these technologies? I want to talk turkey about quantum because we've been talking for five years, we've heard a lot of hyperbole about quantum. We've seen some of your competitors announcing quantum computers in the cloud. I don't know who's using these computers, what kind of work they're being used. How much of the, how real is quantum today? How close are we to having workable true quantum computers? Can you point to any examples of how that technology is being used in the field? So it remains nascent, we'll put it that way. I think part of the challenge is we see this low level technology and of course it was Professor Richard Feynman who first pointed us in this direction more than 30 years ago and I trust his judgment that there's probably some there, there, especially for what he was doing, which is how do we understand and engineer systems at the quantum mechanical level? Well he said a quantum mechanical system is probably the way to go. So understanding that, but still, part of the challenge we see is that people have been working on the low level technology and they're reaching up to wondering will I eventually have a problem that I can solve? And the challenge is you can improve something every single day and if you don't know what the bar is then you don't ever know if you'll be good enough. I think part of the approach that we like to understand can we start with the problem? The thing that we actually want to solve and then figure out what is the bespoke combination of classical supercomputing, advanced AI accelerators, novel quantum capabilities. Can we simulate and design that and we think there's probably nothing better to do that than an exascale supercomputer? Can we simulate and design that bespoke environment, create that digital twin of this environment and then after we've simulated it, we've designed it, we can analyze it, see is it actually advantageous? Because if it's not then we probably should go back to the drawing board and then finally that then becomes the way in which we actually run the quantum mechanical system in this hybrid environment. So it's Nathan, you guys are feeling your way through, you got some moonshot, you work backwards from use cases as a more of a discovery navigational kind of mission piece. I get that. And exascale has been a great role for you guys, congratulations. Has there been strides though in quantum this year? Can you point to what's been, has the needle moved a little bit a lot or I mean it's moving I guess there's been some talk but we haven't really been able to put our finger on what's moving. Like where's the needle moved? I guess in quantum. I think that's part of the conversation that we need to have is how do we measure ourselves? I know at the World Economic Forum Quantum Development Network we had one of our global future councils on the future of quantum computing and I brought in, I tripled a fellow, Paolo Gargini who created the International Technology Roadmap for Semiconductors and I said Paolo could you come in and give us examples? How was the semiconductor community so effective not only at developing the technology but predicting the development of technology so that whether it's an individual deciding if they should change careers or it's a nation state deciding if they should spend a couple billion dollars we have that tool to predict the rate of change and improvement and so I think that's part of what we're hoping by participating we'll bring some of that road mapping skill and technology and understanding so we can make those better reasons investments. Well it's also fun to see super computing this year, look at the bigger picture obviously software, cloud natives running modern applications, infrastructure as code that's happening, you start to see the integration of environments almost like a global distributed operating system that's the way I call it. Silicon and advancements have been a big part of what we see now, merchant silicon but also DPUs are on the scene so the role of silicon is there and also we have supply chain problems so how do you look at that as a chief architect of Hula Packard Labs because not only have to invent the future and dream it up but you got to deal with the realities and the realities are, silicon's great we need more of that, quantum's around the corner but supply chain, how do you solve that what's your thoughts how is HPE looking at silicon innovation and supply chain? And so for us it is really understanding that partnership model and understanding and contributing and so I will do things like I happen to be the systems and architectures chapter editor of the IEEE international roadmap for devices and systems that community that wants to come together and provide that guidance so I'm all about telling the semiconductor and the post semiconductor community okay this is where we need to compute I have a partner in the applications and benchmarking that says this is what we need to compute and when you can predict in the future about where you need to compute what you need to compute you can have a much richer set of conversations because you describe it so well and I think our senior fellow, Nick Dubay he coined the term, internet of workflows where you need to harness everything from the edge device all the way through the exascale computer and beyond and it's not just one sort of static thing it is a very interesting fluid topology I'll use this computer at the edge I'll do this information in the cloud I want to have this in my exascale data center and I still need to provide the tool so that an individual who's making that decision can craft that workflow across all of those different resources and those workflows by the way are complicated now you got services being turned on and off observability is a hot area you got a lot more data in cycle, in flow I mean a lot more action and I think you just hit on another key point for us and part of our research at labs I have as part of my other assignments I helped draft our AI ethics global policies and principles and not only getting some advice about how we should live our lives it also became the basis for our AI research lab at Hewlett Packard Labs because they saw here's a challenge and here's something where I can't actually believe maintain my ethical compliance I need to have engineered new ways of achieving artificial intelligence and so much that comes back to governance over that data and how can we actually create those governance systems and do that out in the open That's the can of words we're going to do a whole segment on that one piece I want to ask you where Rubber meets the road is where you're putting your dollars so you've talked a lot of areas of progress right now where are you putting your dollars right now at Hewlett Packard Labs? Yeah, so I think when I draw my 2030 vision slide for me the first column is about heterogeneous how do we bring all of these novel computational approaches to be able to demonstrate their effectiveness their sustainability and also the productivity that we can drive from them so that's my first column my section column is that edge to exascale workflow that I need to be able to harness all of those computational and data resources I need to be aware of the energy consequence of moving data, of doing computation and find all of that while still maintaining and solving for security and privacy but the last thing and that's one was a how one was aware the last thing is a who and is how do we take that subject matter expert I think of a young engineer starting their career at HPE it'll be very different than my 33 years and part of it is they will be undaunted by any scale they will be cloud natives maybe they're metaverse natives they will demand to design an open cooperative environment so for me thinking about that individual and how do I take those capabilities heterogeneous edge to exascale workflows and then make them productive and for me that's where we were putting our emphasis on those three, when, where and who and making it compatible for the next generation we see the student cluster competition going on over there this is the only show that we cover that we've been to that is from the dorm room to the board room and this super computing now is elevating up into that workflow integration multiple environments cloud, premise, edge metaverse this is like a whole other world and but I think it's the way that regardless of which human pursuit you're in you know everyone is going to be demand simulation and modeling AIML and massive data analytics that's going to be at heart of everything and that's what you see that's what I love about coming here this isn't just the way we're going to do science this is the way we're going to do everything we're going to come by your booth check it out, we've talked to some of the folks HPE, obviously HPE discovered this year Green Lake was center stage it's now consumption as a service for technology whole another ball game congratulations on all this I would say the massive, I wouldn't say pivot but you know a change how you guys operate and you know it's funny sometimes you think about the pivot to as a service is benefitting the customer but as someone who has supported designs over decades that ability to operate at peak efficiency to always keep in perfect operating order and to continuously change while still meeting the customer expectations that actually allows us to deliver innovation to our customers faster than when we are delivering warrantied individual package products Kirk thanks for coming on Paul great conversation here you know the road to quantum is going to be paved through computing, supercomputing software integrated workflows from the dorm room to the board rooms the queue bringing all the action here it's supercomputing 22, I'm John Furrier with Paul Gillan thanks for watching, we'll be right back