ISC 2023 - Quantum Computing

foreign welcome to thecube's coverage of high performance 2023 covering everything HPC machine learning AI high performance analytics and Quantum Computing part of the ISC 2023 Cube event coverage the segment today is quantum Computing next Hardware accelerations right around the corner it's the next next gen coming got two great guests here and a power panel Burns Healy emerging technology researcher with Dell research office and Mike Rebel art senior distinguished engineer with Dell gentlemen thank you for joining me on this Quantum Computing topic yeah thank you for having us John we appreciate it so we we love we love high performance super Computing super cloud super apps the Market's totally primed for this big wave coming and quantums right around the corner some of the smartest people or Engineers are working on Quantum it's quite the buzz and it's legit it's got a whole nother Paradigm Shift HPC professionals and customers they're smart and they're into Hardware they're into all super Computing but they're just now putting their toe in the water with Quantum but they're very aligned with the concept so let's start with kind of just setting the table can you guys describe what Quantum Computing is in a way that the a regular HPC Guru or professional can understand and how it's different and what are the advantages they're going to be yeah I'd say that the best uh mental model for Quantum as it would serve an HPC expert HPC customer somebody interested in HPC will be to think of it as another kind of Hardware accelerator so if you go back and you think about gpus right of course gpus were originally designed for graphics now we're finding applications for them in AIML you know training say large language models other kinds of workloads like that and there are specific workloads that you see that run better on gpus and then other workloads for which they don't really provide an advantage and I would encourage everybody to have that same thought process in terms of quantum so there are going to be particular classes of workloads that Quantum Technologies are going to have the capability to accelerate and really improve you know over our current methods of of processing those workloads and then there are others that they won't so it'll be just a journey of as the technology matures identifying those places where we can gain an advantage and trying to continually average more and better Quantum techniques to expand that that world I mean I like the GPU example because I think people can relate to that I remember when gpus came oh gpus graphics accelerator it's a bolt-on it's a it's a it's a hardware assist specific use case had a beachhead you had one use case now it's everything machine learning similar scope with Quantum it's going to land on some use cases what are some of those things that you see popping out right out of the gate yeah no so the uh the big one that of course captures a lot of attention is when when people say that quantum computers are going to break encryption and and uh it's going to be fundamentally a lawless world where every possible message is now interceptable and decryptable by any you know any actor with an internet connection and uh what is true is that our current encryption standard which is RSA with a specific length key is susceptible to a faster Brute Force attack via quantum computer um but it's very specific to that particular encryption scheme to that RSA uh encryption method so in that realm RSA is one particular workload if you're talking about decryption that quantum computers could be advantageous at but other encryption methods are not susceptible to that same kind of attack so even within domain I think identifying those uh particular workloads that Quantum is helpful for is uh is a bit of a fine line so encryption decryption uh as well as things like Quantum machine learning are coming online I mean that's just like all technology Evolutions you know the previous generation is not as good as the next one and if you don't move to the next one I mean look at the you know everyone's doing two-factor authentication people like that's crazy a decade ago no one did that right I mean well some people did I did but but that's the point and so I think one of the other uh things is is pretty Universal is this idea of optimization and um every company is always trying to optimize their business operations and quantum computers appear to be very well suited uh for uh for optimization types of problems so there are a lot of applications uh it's just a question to go and finding them Michael that's a great point and I like the next the next question I wanted to ask you guys is so many Quantum Technologies out there Quantum Community Technologies what's the map look like and what do you what do you look for in a company working in the area that impresses you what so what's some of the uh um the Napa the landscape look like out there in this market for companies what's who's innovating what does success look like right now I mean I'd say everybody's innovating uh what's what's so beautiful about being so early on in the in the maturation of this technology and so we've got a lot of different paths you know every company has their own road map for scaling up if they're a hardware company scaling up to additional qubits which are the fundamental you know unit of quantum Computing or if they're an algorithm development company they've got all of these different areas that they want to try and tackle and so I see all of these ambitious really excited uh technologists going after this and I think there's a lot up and down the stack a lot to be excited about I'd say yeah what's what's the what's the landscape point of view like are people sharing I mean open source for example everyone's sharing content um what is a what is what is the landscape of the companies look like that impress you what are what what what's um what's the what's the company landscape like like what capabilities yeah so you know we're not going to be picking uh winners at this point right we're on this or on this uh this interview but you know when you see companies that understand the physics when you see companies that are able to uh connect to user needs to user workloads when you see companies that are taking and advancing the engineering for building these systems right that's a great Quantum company and there are a few examples of that uh in the space and you asked a second question around how competitive are these companies uh it's a it's an interesting time right because there there's cooperation uh but there's also competition and you've got uh multiple uh threads of research and development that are relatively independent all kind of you know progressing independently and at this point it's really too early to call yeah and it reminds me of we just we're talking a lot about Ai and other segments around ISC and HPC and how AI has just exploded because of the compute and large-scale cloud and data models that are out there which is a current situation that's what's happening now that wasn't available years ago so when things just explode they just go exponential what kind of problems does a Quantum address in the in the HPC world what are we looking for for that pivot point that that that kick up what what needs to happen what do you guys Envision um the sequence because I'm imagining that Quantum is just gonna it's gonna continue to advance people are going to collaborate coopetition is going to happen but there's going to be a couple bits that have to flip you know yes yes can you share your vision on that because I think that's really fascinating absolutely and and one thing that I think it it uh is helpful to remember is that when we say Quantum Technologies right it actually encompasses a whole lot of different kinds of architectures so uh in our experience right now there's kind of two main competing standards or not even competing standards but complementary standards for what a quantum computer does so there is a a kind of universal gate-based qpu machine so these are the kinds of of machines that you've probably more heard about in terms of what's available uh in the literature out there but also in addition to those kind of universal processors there's a very specialized kind of processor called a Quantum annealer and the quantum annealer uses different parts of of quantum mechanics to do computations and this is what Mike was getting at when he said optimization is a very hot topic and a very uh you know is an area in which we'd expect to see a Quantum at least provide some kind of Advantage hopefully in the near future and these are the kinds of problems that are your sort of large Enterprise scale uh large search based problems that you want to optimize revenue or you want to optimize time or resource allocation and that's one place I'm I'm excited to see coming online soon because there are quantum annealers that are beginning to scale and I think as we get the chance to explore our current optimization problems on those Technologies I think we'll see where that fits in Michael what's your take Renaissance and computes here I mean it's just more Headroom all around yeah and I just want to you know focus in a bit on the second part of your question there the you know is it better than classical HPC uh what are the differences there and and you know I I think we're going to see an evolution of you know what we think HPC is much like we saw in evolution there was a time before HPC had gpus and and that was fine right and and today it has gpus and and that's that's even better and and there will be a future where it will uh be augmented and I think the only question is how much and um you know I think that's the exciting part so you know I'm looking at some of the comments online around orchestration intelligent orchestration um Quantum being another tool in the HBC Arsenal this is a really key Point getting that orchestration of resources is very key um how do you guys react to that what's what's that mean right so I want to think of a workflow when you're you're talking about some large you know workload or some large job or some large sequence of jobs and you imagine now not only do you have possible accelerators in the form of the GPU but you also have Quantum processing units qpus and you also have Quantum annealers and you have other technologies that could provide advantage and so before that that decision was an easy one do I want to run this on a CPU do I want to run it on a GPU right and now you have a much tougher question to answer which is which pieces of these should be of this workload should be allocated to which resources and I think that's where the intelligent orchestration comes into place so it's breaking down those jobs identifying the different components of them and where they fit best in terms of what's going to Advantage them what kind of Hardware architecture is going to match this workload best and it's not just a matter of enhanced execution time but you also have questions of accuracy Fidelity you have availability of Hardware you have price of Hardware so all of these questions get rolled into one kind of final decision and at the end of the day that decision is being made in an HPC environment right the quantum computer is not doing the orchestration that is still a classical machine running on Hardware that we're all used to that is making those decisions and dispatching those parts of your your workload that's a great point in fact my next question was going to be you know in this event in the industry you know the hallway conversations people are saying HPC and Quantum Computing are both necessary to use a quantum computer working together can you explain what that means is it do you agree with that statement and that's going on in the industry and what does it mean yeah I would I would absolutely agree with that I think that uh we should see uh as you mentioned a tool in the HPC Arsenal Quantum is is another accelerator you can't use a GPU without us without a CPU right without a traditional computer behind it um it doesn't make sense and the same thing is going to be true for Quantum uh indefinitely because we're going to need to orchestrate those resources and see what makes sense but especially in the near term we're still in what we in in the quantum landscape like to call the noisy intermediate scale Quantum or nisk era and what that means is while these devices are impressive they still are subject to a lot of stochasticity a lot of noise you typically run a single workload many many times in order to try and drown out that noise with the signal and recover some uh some uh real meaning to your end result and so as we remain in that nisk era in addition to the orchestration you have questions like pre-processing by classical infrastructure post-processing by classical infrastructure you might have to take an output and interpret it you know using your classical Hardware uh so so it's it's still the case especially now that Quantum is going to be able to do a piece of it but it's going to be kind of sandwiched by these pre and post uh classical layers that are going to be needed to interpret what it is that Quantum technology has given you and and just to add to that a little bit um we are always going to be wrapping quantum computers with classical electronics and classical machines they really have a symbiotic relationship and and I think the way we should think about it is that Quantum is going to augment classical and classical enables Quantum and and so you know the ratios may change uh where we run particular inner Loops may change but there are a few significant differences that mean both are incredibly important I mean today we have the ability to do uh uh you know 80 billion transistors in a in a device and we can do it at scale uh we can make incredibly reliable devices and it only takes you know 24 transistors to build an Adder as an example um addition it will be done with classical machines right there are other uh use cases other applications other things that we'll we'll do for Quantum and we could we could break that down a little bit more but I think the key takeaway here is is uh you know these are these are symbiotic Technologies yeah I agree I mean I think if you look at I agree with you 100 if you look at the evolution of what's going on with large scale I mean just take take the hyperscale and then the on-prem and Edge that's developing it's basically distributed computing okay that's the distributed Paradigm that's going to be like a piece I call like the PC it was once the server is now the internet right so everything's distributed you're starting to see the subsystems being more focused the focus on silica and the focus on chips to focus on the connection points between them so again the GPU is a great example so I could see Quantum integrating in with HPC clearly as a beachhead and then from there it's whatever whatever track it takes I mean it has that's just Evolution that's just the computer industry that's how it works yeah and it's it's very different than you know if you look at the earlier phases of compute where uh you know we had vacuum tubes replaced mechanical systems and you know semiconductors replace vacuum tubes um it's not going to be like that right there is no replacement right we will use semiconductors and um and um you know if all goes according to plan we will have Quantum machines and and they will be together it's an operating system it's all the same it's the same Theory I love it anyway I don't want to harp on that but I want to get into some specific Market questions um some industry pundits are predicting a Quantum winter the hype Cycles obviously as into well well into the development is do we are we in the traffic disillusionment what do you guys think about this what's the comment on this quote Quantum winter yes or no I I can't say anything about a Quantum winter I'm not unfortunately a weatherman um but I can say that we need to uh make sure that our messaging is clear from from within you know the quantum industry I think it's our responsibility to make sure that we're accurately representing the state of the technology and accurately representing what its capabilities are right now I think road maps are great and so far uh in my experience companies have by and large lived up to those road maps and I think that if you read the capabilities and you read the road maps and you understand what they're saying you will you'll be seeing that really by and large we're trying to under promise rather than over promise I think that um there I I wouldn't foresee any kind of trough of disillusionment as long as we are understanding where the state of play is and we're not allowing ourselves to get too far ahead of ourselves in terms of of the timeline that we see that the technology maturing at I think the expectation setting is right on I think that's key point I mean if you over play your hand here the tech the scale Mike you just talked about this earlier about the classical intersection with classical computing I think it's more of a journey you know it's it's not about Seasons uh it's a journey and and there is work that we have to do and um you know as long as we're clear and understand that work it will progress so the reality settles in people who work it's not like the industry's not going backwards all right it's going forward um always moving forward so the question on on Quantum what being practical okay some are saying when it reads 1 million qubits that's going to be you know decades before you know weeks or million what do you think about that is that off base how much faster can Quantum scale vis-a-vis say the classical trajectory um is there can you get a feel for what we are where we are in the progress there is there's one particular aspect to this question that makes it both very difficult to answer unfortunately but also gives me a lot of Hope in optimism and so what's really particular about this is that there's so many different kinds of even when you zoom in on your gate-based traditional quantum computer as you might think of it even within that schema there's a lot of different Technologies a lot of different Hardware architectures that are achieving that outcome so you have things like trapped ions you have neutral atoms you have photonics even within the schema of manipulating a qubit and measuring it which is fundamentally all that a qpu is doing there's a lot of different ways in a physical sense to make that happen and so all of these different Technologies are advancing in parallel and already we're seeing some of them scale better some of them have better coherence times some of them are cheaper to manufacture or easier to maintain and they've all got their own advantages and of course disadvantages and what makes me optimistic is that all of these players in the in the field are gonna keep going right they're all going to keep working on their technology going to try to scale it and going to try to achieve uh some kind of Benchmark that they've set out for themselves and what that Benchmark is whether it's a million qubits or it's not a million qubits will depend on the technology too one of the reasons one million qubits has been set out there as this kind of Benchmark is because there's the idea that you're going to need Surplus qubits than you really need in order to do that error correction you're going to need what's called a logical qubit made up out of many physical qubits and you may not need that or you may not need a thousand to one ratio if the Fidelity of your qubits is really high so it's difficult to predict but I'm really optimistic uh about the overall timeline just because there's this there's this diversity there's this growth in the industry and there's starting to be this really really Keen attention and investment paid to it yeah I mean it's a moon shot metric I guess my final question to wrap up here is why should people start paying attention to Quantum now and how HPC is integrated into it this is where we're at now how do I think about it what steps do I take what's in it for me what's for the people in the industry why should people start paying attention to Quantum now so let me let me jump in here for just a moment um you know if we look around right there are plenty of impossible things all around us and those impossible things are the function of a large Journey right uh manufacturing constant Energy Engineering Innovation uh science uh supply chain right and over long periods of time and Quantum will be no different and so what we need to be thinking about is how do we individually and as organizations add energy into this virtuous cycle how do we make the flywheel go faster so that we can all benefit from you know from the the outcomes that are possible and so you know there are folks whose responsibility will be to as Byrne said figure out error correction that's a huge technical problem uh other folks are focusing on trying to uh train up a new generation of Engineers to be able to use this technology uh as businesses we should be focusing on the applications that accelerate our businesses um and and there are people such as yourself John that serve to facilitate the conversation right to build the connections between people so that we can you know all uh head towards this this common goal and so it's really when we think about this we need to think of a journey we need to think of a flywheel we need to figure out how we're going to add energy into this so that we can we can take the benefit burns your thoughts yeah absolutely I I mean one additional answer so everything Mike said is is absolutely correct and we do want to see the industry moving forwards and that can only happen when people are thinking about the technology and there's a lot of upskilling to do there is going to be a little bit of a learning curve when it comes to Quantum not because I think it'll be different difficult to use per se uh but because the kinds of quantum algorithms that we see that are going to provide us with Advantage uh are counter-intuitive they're they're quantum mechanical nature they utilize superposition and it's just a natural not a natural way for us to think so I I'd love to see developers and people in charge of optimizing uh given uh scenarios I'd love to see them start to think about it now and start to wrap their heads around what what is the capability of this technology and it won't it won't be that long until we see we see a piece of Advantage where Quantum is a portion of a larger larger heterogeneous compute environment well I say we start teaching physics in first grade let's get that on the table right away okay so that's number one since everyone's gonna be coding anyway with no code low code just get physics and into the curriculum at first grade why don't we start with there we are so enthusiastic Mike we we are uh into Quantum we believe this is an evolution um and when it's going to hit a Tipping Point and we're gonna we're gonna continue to to amplify the uh the industry we think Quantum is positive uh for the industry and overall everyone in HPC compute super Computing is being redefined super cloud is emerging and super apps are coming onto the scene with AI in there all going to need Computing that's what we see clearly the next 20 years as a shift and we really appreciate you guys doing what you do and thanks for coming on this program just final word put the plug in for Dell what are you guys looking to do for collaboration uh it's a community people are looking to do deals work together share information take the last minute guys and share your thoughts on on what Dell's doing and what you guys are looking for absolutely so and when we say that that Quantum currently and probably always will or will always rely on that classical uh infrastructure layer that's where we want to play we want to play in this organization orchestration optimization layer where we can send the workloads to those appropriate uh resources in those appropriate accelerators and really be that orchestration piece where we're the first Contact for a workload we can break it down do that optimization and orchestrate and allocate these resources as is appropriate for the best you know the best end result for the customer Mike yeah uh you know start by looking at dell.com and the quantum Computing uh page and uh reach out there's some resources there you can learn some things you can meet some people and you can learn about the Dell Quantum Computing solution so that would be my uh my quick plug HPC and Quantum go go better together and that's that direction and thanks for coming on this program what a great panel thanks for unpacking and discussing bringing a Quantum conversation to the table we really appreciate it okay this is thecube's coverage of high performance 2023 part of ISC 2023 event coverage I'm John Furrier your host thanks for watching [Music] thank you

2023-05-28 12:46

Other news