21: Intel's Jim Clarke talks quantum computing
Welcome back to tech to Topi the podcasts by better future John Biggs shall we have Jim Clark works at Intel's quantum computing lab. This is tech topa. Hey, guys, John Biggs here, it's not smart to ignore changes and technology that will eventually change our lives better or worse thing that's not smart, but you know, what is smart going to ZipRecruiter dot com slash techno to hire the right person. ZipRecruiter doesn't depend on candidates finding you. It finds them for you. It's a powerful matching technology. That scans, thousands of resumes identifies people with the right skills, education and experience for your job at actively invites them to lie. You get qualified candidates fast, ZipRecruiter's rated. Number one by employers in the US is writing comes from hiring sites, trust pilot, with over one thousand reviews. And right now, my listeners can try ZipRecruiter for free at this exclusive web address ZipRecruiter dot com. Slash techno at ZipRecruiter dot com. Slash techno ZipRecruiter dot com. Slash techno ZipRecruiter. It is the smartest way to hire. Welcome back to Topi of the podcast about a better future. I'm John Biggs today. We have James Clark director of quantum hardware Intel. Okay. So quantum hardware is probably so my favorite kind of hardware, Jim, why don't you tell me a little, but what you guys are doing there. So if you're familiar with conventional computing, you're familiar with the central processing unit or CPU migra, but Intel were developing the quantum processing unit for bottom computers. Are a quantum computers are an emerging technology on that. The should bring about exponentially vast compute technology compared to to the technology we use today. So what does that? So this is a this is it's important that we get, I guess, are are definitions right as we begin. So what's the primary difference between a quantum computer and a regular computer. So a quantum computer uses to principles of quantum mechanics to access, but an exponentially large space for compute in comparison to classical computer on the way to think about it is is often do with coins. If you have a coin in your hand, it's either heads or tails. If you have a transistor, it's encoded as either a one or zero, the transistor is on or off, but with a Quant of information where cubit will call it a quantum device. It's better to think of that choline is spinning on a table where that coin can simultaneously represent. Both heads and tails list some probability. If you then bring in second bit of information to bit, you can represent a total of four states, three cubits. So imagine three spinning coins. You would have a total of eight states at the same time. What's interesting about this as you keep getting larger and larger number of Cubans, you can represent. Very quickly more states that are acceptable with a classical computer. Or with fifty cubits. You reach a point where you're essentially a bit more powerful than a any supercomputer on us. And the and forty cubits. What's the, I guess is their size difference in terms of in terms of Bill to to generate this stuff is is, is it as big as a regular supercomputer, that kind of thing? Because I think it's, I think it's. Is it my own? You know what I mean? I'm sorry. Can you repeat that? Would I ever Bill to put this in my phone when I have a a? So here's the since two computer. Yes. So here's the comment that I would make in the nineteen sixties when you had just a few different supercomputers on earth or arch computers. They weren't even called supercomputers at the time. That's about the stage that we are in development with the quantum computer. So at first they're really only a few of them, the relatively simple. In the sixties. You could just about check your work. Through a human calculation. That's where we are today. With Christy gonna get to a point where we can do things that no conventional computer on earth to do, but the point at which she would have it in your back pocket. Let's assume that just like a regular computer, let's assume that that's still another forty years away. Okay, interesting. Okay. So the, I'm I'm a big fan of quantum computing and the thing that I'm looking forward to the most is essentially a general access to a quantum computer. I think there's a couple of companies that are working on right now, and I'm sure Intel's doing some more there is that is that something I should be excited about? Is it important for a regular programmers to have access to two machines that can use cubits and things? Yeah, it's all part of an ecosystem that will have to evolve to bring quantum computing to reality programming. Quantum computer is unlike classical computer in the way you manipulate the. Algorithms or the cubits is fundamentally different. So by having these accessible either to small groups at I or eventually to the public, I think you're going to see acceleration of of the quantum computing technology. Now we're in a stage where a we have our our two-bit ships, and we're still continuing with internal testing erica's -ation up prior to putting anything on the cloud. But I think the cloud is probably a good model for early on years in would be perhaps variety of whether it's from companies from national labs. I think you'll see a access through the cloud. And what are we gonna have one. We're gonna have generally available quantum compute. When is it? When is it gonna replace of the regular supercomputers that we have now? Yeah. So a couple of comments there. The first comment is on computers exponentially faster for only certain algorithms at the moment I'm Doug rhythms would be taught. Tiger Affi. Some could be in portfolio optimization. We feel that the first applications will be in chemistry in chemical modeling, which of course, chemistry as a segue into medicine and biology. So in that sense, they won't replace, but they will augments a supercomputer for certain algorithms. So we would expect quantum computer to be stationed next to a supercomputer in a data center. As an example. Now when we will have them, we think that a quantum computer that will say, change your life or mine is still about ten years away up until there will be a lot of learning of it will happen, but it won't be commercially relevant to a company like Intel and probably you won't be able to implement something in your in your life and say, that's because of a year. Okay. Interesting. So what's the in terms of augmenting supercomputer Witter some of the. So you said Jussi portfolio management, I guess network network. Management, that sort of thing is there, is there the idea would be to offload a lot of those functionalities to to to a quantum computer as necessary or or how would that? That's a great point. So for certain types of algorithms are certain tasks a, you would offload them as offload them to a quantum computer as an accelerator. For example, now I expect the number of applications to increase as we have initial systems on, but with the way to think about this as an accelerator in data center. So they so the quantum computer work side by side with the with the other machines it does now. One of the things I wanna point out as part of this discussion is while my team is making a the on processing unit. Other is a host of classical electron IX that would be used for the control system for hosting the the architecture on or a basically that the overall in overall on computer will have on a processing unit than a host of classical compute components. So really any system build is going to be a hybrid system. And then any application we run is likely going to be a hybrid application where parts of it run on a supercomputer or host computer and parts are offloaded to on. Interesting. Yeah, I never never thought about the that you wouldn't be able to use the quantum computer the way it used the generally available supercomputer. Are there any other any current implementations that you think that are fairly cool. Well, cool can be interpreted as a as all of these systems today operated a fraction of a degree above absolute zero. So to that extent, they're all pretty cool. Now, this is an exciting space and. You know, while while Intel is focused on to. Cuba technologies, we certainly it's engine to the field and the people on my team or just excited about any new development myspace. Because collectively with the research all over the world, every big development is one step closer that the overall immunity gets to realizing system. So so it's it's a pretty fun time. Big advancements just about every month. Is there anything particularly interesting that happened recently. So I think I think it until what's what's certainly been interesting is an exciting this year is up until now a lot of the quantum. Processing units or Cuba chips have been built in university settings and university laps. This is where the expertise says. Now, while this is still a fundamental science project. Our view at Intel is if we can take a healthy dose of Intel, engineering and process control something. We're very good at an apply it to. To Cuban technologies than we can help the field exhilarate to that extent. This year we have been running. A specialized quantum fabrication process in our three hundred millimeter fabs in Oregon. So we're seeing. Yield and uniformity on device to device that we feel has exceeded anything that's currently out there. So that's certainly been something that that this ongoing that's exciting here until in. Is this mass production or is this I don't know a couple dozen chips per run? Yes. Great question. So this runs this runs in the same fabrication facility where we do our late stage are in d. So this is where in a in a facility where we run thousands of waivers a week. Now here we're running a few wafers at a time where optimum process when we have a good waiver for that, waiver will have a tens of thousands of two bits allbeit not interconnected. So we have a huge real estate with which to test and play with these devices. And that's pretty exciting. We're producing these on a scale of the hasn't been done before, and one of the challenges is characterizing in in getting as much information as possible from one of these large waivers and feeding it back into our feedback. So they so the expectation now isn't isn't to actively run anything on these things for you guys. It's more just figure out what the what the physical plant is, what the what the physical manufacture of these things looks like in the future. Right. When you say run, I mean, that would be, let's say, for this technology, which this technology is a newer technology. All at the moment is not to create out, say in online system in the cloud attest types of two bits here. What we're trying to do is characterizing some of the fundamental variability of one Cuba type to another. This is a mung, the one of the tasks that we're doing that we're doing now. If I want to become a quantum computer programmer, if I wanna start exploring this this space, which would I start doing right now? And that's a great question. We get asked this question a lot. How do I get into Guam computing? In fact, this topic came up last week. I at the White House, the White House had a quantum information summit in workforce. Came up as is a topic for discussion. The answer is you don't have to be a quantum physicist to work on quantum computers on my team. The team at until we have people that are materials experts. We have people that are transistor experts. We have people that are f microwave engineers with the signaling, and then we have people that are architects who are trying to apply Intel architecture more to the quantum space. So certainly there is a component of of. Knowing. I'll say, quantum mechanics that would help, but it's far from the case that you have to be a quantum physicist to to work in space. So you don't have to be on business, but you would what it does. Is there any sort of specific programming language as I should be looking at as a as a student or whatever, or is there or is the, I guess the question is, do you have to be super deep into quantum computing to be in quantum computing? Yeah, that's that's our point. I'll answer two ways. One is I think most of us would say, you don't have to, but that that there are a lot of areas of expertise, which when applied to quantum computing help exceleron field. The second is I think that universities now are grappling with what would a quantum information systems degree look like. It's probably a does of Bonham physics and computer engineering, classical computer engineering. So I think this is an area is an area of particular importance because there simply isn't enough, aren't enough quantum engineers around today to fill the openings both in academe Lia in industry. So we need to develop more of them. Very cool. All right. So at what point is a quantum computer gonna be able to run every single car in a an a self-driving city. Yeah. Under great question. An area of of of work with quantum computers now is optimizations. Imagine if you have a self driving parts looking for the optimal route. There's also a strong on machine learning component which we're seeing. Abors with with quantum computers. So. Let's see. I'm not an expert on when the first self driving cars will come. We were to say that each of those is still in their infancy than I would expect a still, at least on the ten year timeframe before on computer are influencing the self-driving our space, but it's certainly an active area of vision if not already in the research. Okay. So about about a decade before we before we see it really take off that sounds far to you sounds far off. It really isn't. If we take a look at the inflection technology, the first transistor was in the late forties. The first integrated circuit was in the late fifties, and the first micro processor was around nineteen seventy. And so you really fundamental changes in technology about every ten or eleven years. So when I say ten years away, that's not so far away that we shouldn't be excited about it or art boards. I think the the frustrating thing when I do this podcast, I talked to a lot of folks who are thinking about the future, and I always say that we're sort of an interstitial period right now. We're not. It's not neither here nor there. We don't have. We don't have a mobile giant. We don't have a mobile explosion right now. Everything's kind of flat over there. We don't have any microprocessors. It can't get much the traditional microprocessor can't get much better from what it is right now, which is why I'm particularly excited about generally available quantum. But let's see. Let's see how that, how that all plays out. So one of the misconceptions that I hear off to your point is here's someone at a conference or even at the White House event last week who will say Moore's laws ending, therefore, we need to work on quantum computing. Ability and you wouldn't. You wouldn't expect a different response from Intel. We think Moore's law is healthy and in fact, at on computing person here, I would say that we need the technical advances of a few more generations of Moore's law and order to even bring quantum computing to reality. So if Moore's law were dead today, it isn't a than I was the prospects for quantum computers with also. Interesting. Very exciting. All right. So sue people wanna try some try some cubits. Can they just call you and you can send them a cubit processors, something we're working on that we're getting. We're getting to that point. We don't until we don't make the refrigeration units. So that would be something that they would have to provide getting. I gotta I gotta. I gotta I gotta Amanda downstairs. We just put it in the freezer, right? Not the right temperature. Now those believe it or not, though that's not the hardest part people will say, have the getting to a fraction of a degree above absolute zero as prohibitive there. Many companies around the world that built these refrigeration units they're about is of a fifty five gallon drum. Okay. So. All right. Well, fascinating. This is this is great stuff. I'm excited to see. Next, Jim Clark Intel's quantum computing. This has been great stuff. Thank you. Technophobe is brought to you by happy, fun Corp, happy, fun. Corpuz designed driven technology company in Brooklyn, New York that specializes building mobile and web applications for startups and fortune five hundred companies, whether it's a new mobile web application that will help people experienced the internet in fun new way or software that will interface with a new piece of top secret hardware. Happy fund for is always up to the challenge. Bigger small, have you loves building software and loves working with great people come build with them, happy fund for dot com. Topi is also sponsored by jaywalk. Jaywalk is a new app that pays you to walk. You can try it out at jaywalk. Dot me. It's created by me, John Biggs and a few of my friends. Jaywalk dot me. Please check it out. The.