 Welcome back everyone. I hope you had fun with those farting bags. And if you're just joining us, a special welcome to you for this virtual Super Science Saturday, streaming live from our homes and our facilities at the National Center for Atmospheric Research, or NCAR. Now NCAR doesn't only have research labs in Boulder, Colorado, and in this next segment we're going to get the chance to take an inside look at the NCAR Wyoming Super Computer Center, or NWSC in Cheyenne, Wyoming. Super computers are actually really helpful for us as scientists and the science that we do at NCAR, because it allows us to deal with really complex models as well as lots and lots and lots of information that we can use to study the earth system. Again, we love hearing your science questions. So please submit anything that comes to your mind as we talk about super computers using either Slido, and I'll pop the link in the chat right now. Or if you are in Zoom, you can go ahead and use the Zoom Q&A. And with that, I'll turn it over to Summer to give us a tour of the NWSC. Hey, Summer. Hi, Dan. Thank you. My name is Summer Wasson, and I am the Education and Multimedia Specialist for the NCAR Wyoming Super Computing Center. And so I'm really thrilled to be here today and be part of the Boulder, Colorado Mesa Lab Super Science Saturday. So I'm going to just do a screen share here. We have a couple of things that I want to show you today. The first being, we also have a Super Science Saturday in Cheyenne, Wyoming, which is where our data center is located. Ours has also been virtual in the last year or two. And so I guess if you're in the Wyoming area, keep an eye out on our social media for if when we're able to open and do those in person again. But I want to talk to you about our Super Computing Center. And so as Dan mentioned, we do have a large facility. We have a large supercomputer. And if you didn't know this, there are over 500 supercomputers in the world. Our supercomputer is one of the top 100. And that's really notable. We're located in Cheyenne, Wyoming, and I'm going to give you a quick tour of the facility. So our scientists, you've learned a lot about our scientists and the kind of work that they do and the data that they collect. And we need some way to process that. So that's where our data center comes into play. Our data center is, as I had mentioned, located in Cheyenne, Wyoming, and we have a couple of partnerships that I do want to mention that brought us here. And that's a partnership we're funded mainly through the National Science Foundation. So those are your tax dollars, hard at work. And we're also, we also have a partnership with the University of Wyoming, the State of Wyoming, Cheyenne Leeds, the Wyoming Business Council, and Black Hills Energy. So that's what brought us here to Cheyenne, Wyoming. We also use the climate in Cheyenne to help us cool the facility in a few ways that save us money and are a natural way to cool some of the heat from the data center. So I'm going to show you those here in a moment. This, what we're looking at on this screen is an overview of what our campus looks like. And the square area here is where we're going to go inside. That is where we have our current supercomputer called Cheyenne. We also have a new supercomputer coming online in 2022 named derecho. And if you didn't know, derecho is a long string of storms and the way that they form. And we'll talk more about that when we get into the data center. At the NWSC, which is the NCAR Wyoming Supercomputing Center, we do have a visitor center as well. We're currently not open to the public, but when we do reopen, we'll have that posted on our social media as well. So I want to talk about, before we get into the data center, some of the things that you encounter when you're running a data center in general, and it's kind of the nature of compute. So if you've ever used your laptop at home and you put your laptop on your lap, or you've got it on a blanket on your lap, you'll know that it gets very hot. The same is true with any kind of data center components, servers, routers, a supercomputer. So we need to get rid of that heat in some way, and that we need to get rid of that heat in some way. And so the best way to do that is to try to reduce costs in any way possible. So again, we use our climate Wyoming to help us do that. But most data centers have that same struggle where they have to disperse that heat. It's kind of a byproduct of all the work you're trying to do, all of this heat coming from the computers. And every time a one switches to a zero and a zero switches to a one, that generates heat. And again, that is the process of compute. So most data centers are spending a good portion of their money on trying to cool their facility. You can see here, typical data center spends about 15% trying to do that. And you want to spend more of your budget on it. So at our location, we built our facility in such a way that we're kind of letting natural processes disperse that heat for us. And then we reuse that heat before it is cooled again. So we'll show you how that works here in our data center. I'm going to switch screens here for just a moment. And I want to show you, if you haven't seen it yet in our presentations today, we did create a virtual 360 app. And the virtual visit will show you the Mesa lab, which is where Dan and some of the other folks are demonstrating their scientific work and information today. And we also have the virtual visit for NCAR, which is or the NWSC, which is the NCAR Wyoming Super Computing Center. So I'm going to go ahead and click into the NWSC. And we're going to just take a quick look around here. This is our facility. We saw kind of a quick overview of that when we were in our slideshow just now. But when you visit this virtual visit app, there are hotspots that will tell you a little bit more about the work that we do, the partnerships that we have. And it will also feature some very interesting projects and labs that you can do at home. So inside our visitor center, we have an assortment of kiosks and informational components. Again, we're not open to the public right now. But you can definitely go through this virtual visit app and kind of get a feel for what our visitor center entails. We have also linked to some of the labs. And we've noted the grade levels that it would be appropriate for if you wanted to try some of those at home. And those activities can also be found on the SCIED website, which is scied.ukar.edu. So here's where I want to start our tour. And it's at the end of the visitor center. We are going to come to the glass here and kind of look into the data center from our visitor center. And I'm just going to pan around a little bit slowly so I don't make anybody see sick. But this is what it looks like when you start the tour. And the reason that I start here is because this is one of those components I was talking about the way that we built our data center. We're using a couple of natural processes to help us disperse that heat naturally. And therefore we can spend a majority of our funding on IT and compute interests. So that means we can have more money to buy more computers and do more processing, right? So here we're looking into the data center. And if you haven't been into a data center before, I'm going to zoom in a little bit here. You can see that there are tiles on the floor and those tiles in most data centers, it has a little bit of a two foot gap underneath. And that's where you can run a lot of your cables. Keep the cables out of the way, do a lot of your infrastructure that way. And that little crawl space, that little two foot space is where you would do most of that. In our data center, ours is 10 feet tall. And I want to talk about why that is so important. It's incredibly unique for a data center because you need to have a strong enough floor to carry the weight of a supercomputer and all of those data components. And so what we're doing in this room, all of the heat from the supercomputer from the servers and the network storage, that heat is actually coming up into the ceiling here. It's rising into this corridor and it's being cooled and humidified. Humidified because our climate Wyoming is a little bit dry. But that warm air comes down into this corridor, into this filter wall. And behind the filter wall is a huge fan wall that's that then pulls that warm air through, cools it, and then it comes back up into some graded slots in the flooring of the data center. So that's definitely a unique feature of our data center. And again, it's simple, but it's what helps us cool that warm air from that data center room. Again, that pie chart I referenced at the beginning, that's what keeps our costs down. So this is Supercomputer Cheyenne. Supercomputer Cheyenne is currently ranked, I believe, 60th in the world out of over 500. So that's pretty notable, right? I mean, in Wyoming, Cheyenne, Wyoming, we have one of the top 100 supercomputers in the world. So let me talk about a supercomputer and the way that they are configured and the components that are inside. A supercomputer does not have a hard drive per se. It has processors and it has memory. It kind of offloads all of that data that it processes to another set of servers. And I'll talk about those in a moment. But a supercomputer is really unique. And the way that they process is called parallel processing. So I'll give you this as an example. If you were to go out, find a very large, rural field and you and your friends are going to mow that lawn. But you're going to start out by yourself. You're going to mow this lawn by yourself. It's going to take you probably all day with push lawn mower. And parallel processing is you gather a bunch of your friends and they come mow a strip of lawn next to you as you go. So in your group effort, you're accomplishing the same task. You're doing it much quicker and more efficiently. So that's exactly what parallel processing is in the compute world. Parallel processing is all of the processors in the supercomputer are tied together. They need to know where they are in the process of solving the math problem. And so they have to talk to each other at a high rate of speed, which can't have any kind of latency. So that's called network fabric. So the way that all of those processors are configured together is through a network fabric. So let me just give you some stats about Cheyenne. Cheyenne is a 5.3 petaflop machine. What is a flop? Okay, let's start there. A flop is a floating point operation per second. So essentially Cheyenne is doing a lot of very heavy decimal decimal points math math problems. And so a petaf is a quadrillion. So Cheyenne is doing 5.3 quadrillion math calculations per second. Now that's happening through the parallel processing, right with all of the processors being tied together and being able to work on that math problem simultaneously. So it's a 5.3 petaflop machine. Now it's got 313 terabytes of RAM. If you kind of compare this to maybe your home computer, your home laptop, your home laptop might have a dual core processor and it might have, oh, maybe eight gigabytes or eight megabytes, maybe, or I'm sorry, eight gigabytes or 16 gigabytes of RAM. But so you can imagine how many laptops, how many cell phones, how many desktop computers it would take to create a similar super computer on your own. So Cheyenne is a very fast high processing machine. This is actually you mentioned quadrillion. That's one with how many zeros after it. Like that's a really big number, right? Yes, 5.3 quadrillion per second. Yes. Wow. A lot of computations that can happen at once. Yes, absolutely. Cheyenne is the second super computer to be located in our facility. The first was super computer Yellowstone and a super computer lives for about five years. So due to the heavy processing nature of a super computer versus a normal machine, they start to experience somewhere in terror. They need some maintenance after about year five. So I compare it to this. You have a Ferrari. You've just purchased this Ferrari and you've taken it off of the dealer's parking lot and you get on the highway and you slam down that gas and you leave it there for five years straight. That's exactly what a super computer is doing. So a super computer Yellowstone was a 1.5 petaflop machine. And so in our most recent super computer, we've gotten about three times faster. The cost was about the same. We received funding from the National Science Foundation for about $30 million. For Cheyenne, we also purchased Yellowstone for $30 million at the time as well. So for about the same amount of money, we got three times the processing power. I do want to mention that super computer Yellowstone, we do have a rack that we have provided to the State of Wyoming State Museum. So if you're ever in the Cheyenne Wyoming area, you can stop by and take a peek at what a super computer rack looks like. In this virtual visit app, you'll find some hot spots that talk about Yellowstone and we'll give you a little bit of a tour of the networking fabric that tied all of the components of Yellowstone together. So even though those two super computers were the first two to be located in our Cheyenne Wyoming, NCAR Wyoming Super Computing Center, they're not the first super computers that NCAR has ever had. Our first super computer came online in 1966. And we've had, I believe Cheyenne makes it the 42nd super computer in our organization. So we are not new to the super computing world. We were actually one of the first and in the forefront of using it for scientific and atmospheric research data. In this slide, you'll see a cray. A cray is one of the first super computers that were built in the United States. And the funny thing about the cray is it's shaped in this very small little C shape configuration here. And the funny thing about it is that the reason it was shaped that way is because that was the maximum length of network cable you could have at the time before you introduced latency between the nodes. And latency really throws off that math problem calculation. So you can't have latency when you're doing super computing and you have that interconnect network fabric. And as you can see today, Cheyenne is much larger. I'm just going to open our augmented reality view of Cheyenne. You can see that it's much larger. Due to the invention of fiber optic network cable, we can definitely build out bigger machines with faster processing, faster communication between those nodes. And so we're not restricted to that small little C shape any longer. Also, in the virtual visit app, you can build your own Cheyenne. So if there's anyone interested, the virtual visit will show you how to build what we call a PyAn instead of a Cheyenne. So it's a play on words because you're using Raspberry Pis to build together your own supercomputer and help them process weather modeling. So you'll be able to implement the Worf modeling system and be able to do some of your own weather forecasting using a Raspberry Pi. So definitely take a peek at that if you visit our virtual visit app. So when the supercomputer processes this data, it needs to store it somewhere. And that's what we're looking at right now. This is network storage. And this is really common in a data center today. These pods are using what's called hot aisle containment. So I mentioned that we need to get rid of that heat so that we can reduce our costs and spend more money on compute. So the heat actually from these servers rises up into this corridor, and they're all enclosed in this plexiglass container, which again is called hot aisle containment. The heat rises up into that corridor, goes down and gets humidified in that first scene that I took you through, and then it comes through the filter wall and gets cooled by this set of fan walls and is pulled back up through the floor here. So that's how we disperse the heat. Again, we're trying to just reduce costs so that we can focus more on our compute power. But let me kind of take you into one of those hot aisle containment. Let's talk about why we have to disperse that heat. So here we are inside the hot aisle. And this is what the inside looks like. The backside of all of the servers, you have network ports, video ports, you have an assortment of ways that these components are connected to each other and can be worked on. This is where you would come in to do some of your server system admin network admin work. But let's take a peek here. This hot aisle containment in particular is at 100 degrees, and that is with the air cooling process happening. So as you can imagine, the supercomputer gives off a lot of heat. The servers that it's connected to gives off a lot of heat as well. And again, we just have to be mindful of how we disperse that heat so that we can focus on compute. So I'm going to take us back out here to our network storage. Let's take a peek at that. So I mentioned the supercomputer doesn't have hard drives, it has processors, and it has memory. So when it works on a project, it saves pieces of that project to the direct data network storage that we have attached here. And these are configured in a RAID configuration so that and that's common in a data center. The RAID configuration will help save data in case you have a disk loss. And so the supercomputer saves all of that information to the network storage. And then our scientists can create models and do a lot of their advanced work from there. I do want to go back a screen here because I want to ask if anybody has an idea, and the other panelists can chime in, Dan, if you're still there, how do you think our supercomputer is cooled besides airflow? Ooh, that's a hard one. Does it have anything like AC that we have on a lot of our office buildings and school buildings? A little bit. So our supercomputer, this is very unique, and our new supercomputer to Rachel will be the same. But our supercomputer is a little bit different from the data network storage I just showed you. So the data network storage is using that airflow that's coming through the floor vents and up through the corridor. But the supercomputer is using airflow within itself, in addition to water cooling. So there's actually 65 degree chilled water being pumped through the supercomputer. And it's going straight to the processor chips. So let me just show you here. When you open one of the servers, which is a blade within the supercomputer cabinet, it opens up like a pizza box. And that pizza box reveals the memory and the processors on the motherboard for that, for that blade. So each blade within Cheyenne is running that 65 degree chilled water and it's acting like a heat sink. Rather than having those small aluminum finned heat sinks that you would see on your normal computer at home or on your motherboard. This is a closed copper plate, which is running that 65 degree chilled water. So through thermodynamics, it's actually just kind of absorbing that heat, taking it back outside so that it can be recooled. And we're again using our Wyoming climate to help us cool that water in almost what's considered a free cooling kind of manner. So we have a water heat exchange units that help us do that. We're using the climate here in Wyoming to our advantage. And that 65 degree chilled water is being cooled and processed through our supercomputer to help keep it cool. So I did want to keep a few moments open for questions. Yeah, thanks Summer. So it sounds like a supercomputer is kind of a bunch of computers that are working together to do something. And Violet's wondering, what do people do with a supercomputer? Oh, absolutely. So we have a website called HPCTV. And the interesting part is we, I mentioned at the beginning of this tour that we have a variety of partnerships, right? We have the National Science Foundation funding, State of Wyoming, University of Wyoming. That's kind of a small grouping of people who are using our supercomputer in particular. Because we are part of the University Corporation of Atmospheric Research. That means that we are a collaborative of universities that are running projects. And we have scientists who are also running projects, researchers. And HPCTV will show you in real time the projects that are being run on the supercomputer today. So these allocations can be looked at here. You can also see the specs of the machine, which is really neat as well. It talks about how many petaflops and how many iPhone 7s, that's compared to, we're now on iPhone 11. But let's take a peek at the active projects. Here we can see who is running that project, the organization, the purpose of that project. And this is really insightful to understand the scope and the variety of scientific work that's being done and used in our supercomputer. Yeah, it's pretty neat. I mean, there are folks from all over the world that are gaining access to these computers. I think we have time for maybe one more question. Nicholas is wondering, what's currently considered the top supercomputer in the world? Oh boy, that is a good question. Let me look it up real quick. And I'm going to share this on my screen as well. We got some research and action here. Yes, indeed. Let me stop and do a new share for you. And while you're looking at it, Summer, I'm going to pop the link for folks into the Zoom chat that has both the MWSC and the Mesa Lab virtual tours. So once Super Science Saturday is over, y'all can explore those at your leisure. Perfect. Thank you. All right. So there's a website called top500.org that shows you the other supercomputers in the world. And if you come to lists in top500, you'll see the most recent report in where the supercomputer is currently sitting today and compared to other supercomputers in the world. So what's interesting is in 2017 when Cheyenne came online, I believe it was first ranked 13th fastest in the world. Now, I haven't really talked about derecho much. Derecho is our new supercomputer that will be coming on in 2022. It's going to be a 19 petaflop machine. Currently, we're at 5.3, right? So it's going to be doing 19 quadrillion math calculations per second. And so we've even gotten a little bit better in what we were able to purchase this time. We have gone just a little bit above three times faster than we currently are. But let's take a peek at where things are currently sitting in the top 500. And if you come down to the list here, actually, there is a page to view all 100. So currently, rank number one is supercomputer Fugaku, which is in Japan. And it's the Reichen Center for Computational Science. But you'll see Summit from Oak Ridge National Laboratory, Sierra, which is a DoD machine with the United States. But you can see the difference here. And so this one is 53 petaflops. So there's a little bit of a difference there. And I'm curious to see where derecho will land when we come online as well. But this website is very interesting. It also has the top green supercomputers and data centers in the world. So definitely something to check out. Wow, thanks for sharing all this information with us, Summer. This was really, really cool. Now we'll have to wrap up in preparation for our next show. We do have a survey on Slido. If y'all are interested in providing feedback about this event, I'm going to pop that into the chat. And I don't know about y'all, but I'm definitely getting a little hungry. So be sure to tune in at the top of the hour, 11 o'clock Mountain Time. And we're going to go into the kitchen with Chef Nancy. So I hope to see you then. See ya. Thanks again, Summer.