 the Qualcomm booth here at Embedded World, Nuremberg, and who are you? Good morning, so my name is Leon Pharoasati, I'm Director of Product Management, I represent Snapdragon for Embedded Computing. And this is where we take our Snapdragon processors and we enable and prepare them for going into a number of embedded applications. So let's go over to your Snapdragon wall over here. You have, for example, a whole bunch of Snapdragon 820 and it can go on this small board here. Yeah, that's very impressive. How can it go so small? So, you know, we come from the mobile world. So in mobile, you know, you have small size, you have low power, because these are, you know, arm based devices, and everything is really optimized as far as size and power. So you're getting a lot of performance. For example, in the case of 820, you're talking about four very powerful cores, 64 bit cores, very powerful GPU, very powerful DSP, all in a very small processor. It's not quite the 835, but it's the most powerful embedded chip in the world, no? Absolutely, absolutely. Performance per watt. Intel is doing, right? They will probably overheat. Performance per watt, you know, we've really done optimization at a system level. So this is where you take, you know, everything, including your CPU, GPU, DSP, your video core, your audio core, including connectivity. So integrated Wi-Fi, Bluetooth, GPS, everything is optimized as a whole. That's why you can get so much power, so much performance in a micro song, system and module. With four gigabyte RAM also. Absolutely, yes. Where's the RAM? There's no space. The RAM is actually, no, in this case, the RAM is packaged on packaged. So this is what they call POP memory. So the memory is stacked on top of the processor, yes. So what you see actually here is the memory chip that's stacked on top of the processor. That's fantastic. What can this be used for? You can put this in the... Oh, just, I mean, what do you see now? These modules are going into, you know, digital signage, robotics, medical imaging applications, in-flight entertainment. So, I mean, just a number of things that you can do. Basically, the idea is that you're adding intelligence at the edge. So, for example, you can have a drone that's just taking some dumb imagery, or you can have a drone that's doing the processing and only sending relevant information. You have another example here in the booth where you have a baby monitor. And I think it's... Maybe it's one of those. You're going to see a lot of... Yeah. There's another one here from Altia Systems. This is... You have the business guys over there. You need to persuade them. Is that the 820? It's 20 base. Right here, exactly. Can you go? Yeah. So, right, is that the one we were... This is the one from e-info chips, one of our technology providers. Let's look around it. So can you explain what's here? How do you connect this to stuff? So most of these modules, they have onboard connectors and the idea is that everything you need is on this module and then you have your own system board where this plugs into. So all you need on your system board is mating connectors where this is going to plug into. And you see an example of that. And that's POP, and everything. Yeah, this one is... What are those chips around here? The other chips typically is you have a Wi-Fi and Bluetooth front end. You also have onboard storage. In the case of 820, you can have EMMC or you can have UFS as onboard storage. And you also, every one of our processors comes with its own power management IC. So one of the chips that you see on there is your PIMIC, which does all the power management. Is there any chance you could put this all on the SOC? Well, the PIMIC and the Wi-Fi, Bluetooth, these are analog. So they have to be separate. Yes, everything that's baseband, everything that's digital is integrated into the SOC. You can't make a chip that has both and only one chip and that's it. Well, you can't. Because of course you need to have RF front ends and you need to have analog. So that was not the Aragon, that's another one. This is actually Aragon from A4. So here we're showing, this is a company called Altia Systems. They have created an embedded vision device demo based on the... Can I interview you? Actually, you can interview me. Oh sure. So who are you? I'm Orang Zeppan. I'm the founder and CEO of Altia Systems. We're a California based startup. Building the world's first real-time synchronized multi-camera system on the Qualcomm A20. Synchronized multi-camera. Right. So there are two miniaturized video cameras here. They're exactly the same like the smartphone. They're tiny. Yeah. And we are synchronizing them to create a 130 degree 5.5 megapixel field of view. And you're seeing that on the screen. 120. 130. 130. 130, 5.5 megapixel. And how does it merge in the middle? So that's the magic of our real-time stitching technology. It's keeping my head inside the... Yes. It understands heads and it's trying to go around your head. And you notice it's real time. Yes. So is there enough performance in the chip so you can do all this? Yes. Yeah, it's a great chip. It's got all the right ingredients for us. Real-time video actually pushes performance very hard. So we use about 70% of the GPU and about 30% of the CPU in an optimized pipeline. CL2.0 pipeline. So this hard looks, it's very well-merged in the middle. It's like smart merge, right? Yeah, exactly. It's dynamic stitching. So it processes 180 million pixels per second in the software pipeline to figure out where to put the boundary. So you're giving them a lot of pixels. Yeah. That's right. Are you working with the PanaCast or is it different? Yeah, this is our product. This is our commercial product. This has three in there. Yeah, this is the PanaCast 2.0. It's in high volume production and there's three running at 180 degrees. This one is based on FPGA and here we are generating the scalable pipeline in software only. So this FPGA is not Qualcomm yet, right? This is not. But you could make a PanaCast Qualcomm, right? Of course, yes. Because you're showing it off here. And this can go into a lot of the modern applications for intelligent vision, right? Because we can do immersive video, real-time sensing and connect through cellular or Wi-Fi to the cloud to big data. So we can, you know, see what's coming, understand what's coming and make decisions on what's coming. So this is AI? It is. Partially AI. It's capable of sensing and learning. It understands objects. Right now, for example, it understands faces. So it'll understand people's faces. It'll give you count, location. And it can generally understand objects. And so what's it called? Could you do a whole globe? Yes, yes, of course. You could do a 360. Merge a few chips to do that, though. You can do everything in one chip. One. One chip you can do a 360, 4K. Yeah. But not 4K, 60, right? Not 60, 30 right now. 30. Yeah. And you know, with more technology than, of course. How many lenses do you need for 4K? Three. Three. For the whole, for the whole, all right. So there's a lot of work for you to do, right? Yes. There's a lot of interest. Yes, we're getting a lot of interest. The first time being at this show, you know, we're selling into business applications. And now there's a lot of interest in industrial use cases. So, you know, cars, drones, people like that have come by trains, highways, smart city infrastructure. All of those kind of use cases. All right. Thanks a lot. So let's go back to the Snapdragon wall that you have over there. Yes, let's finish over there. Thanks. Because that was just the first board. But you have like 80 boards right here. So it's going to be, so is this also 80? Yeah. These are, so this one is from Ethan4chips. That's the one. There's another song from our technology provider, Enforce Computing. And of course, you have the option of having a single board computer where everything is already on the board. You don't have to have your kind of system board. If you want a single board computer, for example, you could take something like a SBC. So now this has pretty much everything on board. So, you know, you could put an enclosure around this. And that could be your final product. So these are all production-ready boards or modules. It supports Android. But that's for Adreno. So that enables you to have every Linux distribution. Yeah, one of, so our focus for embedded, you know, of course we have Android coming from mobile. But our focus is Linux. And we are working with Linaro, which is an open source Linux organization, to upstream to the mainline kernel. And one of the things that has also been made available as part of this community enablement is the Fridrino driver, which is, you know, open sources, you know, access to the GPU. So now with, you know, Snapdragon, you can go on developer.quacom.com, where you can get access to software images, whether it's Android, whether it's Linux. Even in the case of Snapdragon 410, we have Windows 10 IoT. Wow. Did they use Fridrino to enable that? Windows 10 IoT, no. They have a special partnership, right? It has its own drivers and it has its own. But everything is now available online. So this is without requiring any licensing or anything like that. But Fridrino... And then there's a Dragonboard. Let's not forget the Dragonboard. Which has like 12 Linux distros that work on it. That's about anything you can imagine, yes. So the beauty of Dragonboard, and I think we have one up here somewhere, right here. So beauty of Dragonboard is Dragonboard is based on open hardware stand-in called 96 boards. Which is the idea that you have these two connectors. This is a low-speed connector. This is a high-speed connector. And it's standardized for that. So now you have all these companies that are creating these mezzanine cards. That are compatible with 96 boards. And Dragonboard is not the only board with 96 boards. There's multiple, you know, other vendors, SOC vendors, that have created 96 boards, compatible community boards. So you have now sensor boards. You have a company called Gumsticks. They've created a drone kit that goes on top of the Dragonboard. So, you know, now it allows you to do quick prototyping. And then in Arrow Connect, I saw some early prototype of the 820. But officially you can't say what's happening with that. I can tell you that so there's a beta edition of the 820 Dragonboard that we made available. What we want to do is we want to start the Linux development process. So Lenaro, who is the organization, we work for Linux Enablement. We're working with their developers. And we're giving them these beta edition Dragonboards so they can start the software enablement. They can start the upstreaming effort. So yeah, I can tell you that the first versions, the beta edition of Dragonboard is available. If you are a developer, if you are part of the Linux community, we'll be happy to work with you and give you one of the beta editions. Yeah, because I saw Rob Clark, who was working on the 820. He was just showing Fredrino working on the 820. That means there's going to be lots of very advanced Linux support. Absolutely. Yeah, I mean the GPU, the graphics core subsystem in the 820 is phenomenal. I mean the power that's in there outside of just the CPU subsystem, the GPU and then the DSP, I mean that combined, now you have tremendous power in an ARM based system. And the DSP is able to do lots of AI kind of stuff, right? Absolutely. That's what they were doing over there. But is that open source? Do they have access to that DSP? Today you have what we call a hexagon. Hexagon is the name of our DSP. You have an SDK. So as a developer you can apply and get access to the SDK. We're working on opening up more and more. So I mean this is a lot of these things we're just getting started on. So the DSP today, you have an SDK and over time we're going to do, open it up more and more. But the work Rob Clark has been doing with Fredrino, it sounds like an amazing kind of work to reverse engineer. Is there any chance that you would actually just say, hey, here's the source for our drivers. And maybe there's a different performance, right? Maybe the official drivers have a little bit more. I wouldn't be able to comment on that. You can comment on that. But who knows? Then you have more demos right here. What is this one? So this is actually one of the first 820 based. This is from one of our other technology providers, Intrinsic. They also have a song. So this is a 820 module. They put a little passive sync on it to get more performance out. And this is a complete board with many, many connectors on it. So this allows you to again do a lot of attached different peripherals and different type of components. This is another super small. This is that song that's on that board. This is the Intrinsic song, 820 based song. Why do you have a heat sync? Is it because they want to use peak performance? Yeah, they want to use peak performance. If you want to get more, you know, if you want to run basically all the CPUs, maximum frequency, then you can add a fancy. Another one I wanted to point out, this is one of the newer boards from one of our other partners, Inaminds. Here what they did is they integrated WM, so basically cellular data on the Snapdragon 410. Is there a SIM card slot? Yeah, this one gives you a SIM card slot and now you can have integrated LTE and 3G on this song. So this is one of the new things I think. This is the first time we're showing this product. Because you're showing over here you have LTE for ILT. But this is just a bunch of models. These are standalone modules. These are where you want to add connectivity to a system that you have. In the case of the Kiteboard, you have now Snapdragon with integrated cellular connectivity. So that's a new one that we're showing here. And then up there you're seeing all the modules and boards based on the Snapdragon 410. I have to also add one of the keystones on the embedded program is the longevity. So we made a press release last year where we announced Snapdragon 410 and 600 so far. That's where you see the E, 410 E and 600 E. We've announced our commitment to make these processors available till 2025 or longer. So that gives the vendors, including the song vendors and end customers, really that extended life. So if you're going into an industrial application, if you're going into, say, a medical device where it takes sometimes two to three years before the device can even come out to market, they now have the assurance that the system on chip, the SOC, is going to be available till 2025. So C is for community and E is for extended life? Or embedded or extended life. For me, embedded and extended life really go hand in hand. Because in order to be a truly and embedded processor, you need to have that extended life. You need to be able to, you know, if you're designing something, we don't want you to have to be forced to change your processor to three years down the road. So we're giving them that 10-year longevity. Let's see, look at some of the other boards you have here. Actually, there's a few boards here also being shown for the first time. These are from some local board vendors, you know, Kaifeng Club, a German company. And they put a 410 on here. Yeah, let's say it's a 410 SOM, but it's a development board with a lot of industrial interfaces that they've broken out. So they have these two different 4-factors using the same SOM, but they have two different industrial 4-factors. We have another also local board vendors CODL. This is their 410-based, 410-E-based system. And of course, you already saw Intrinsic and E-Info chips. Here's another company. Also came out with their, you know, fairly new board. Genie Attack. They have a SOM and a single board computer. Nice, they have a booth here, so I'm going to try to find them. So this one has... What we're trying to show with this wall is the ecosystem of hardware and software around Snapdragon is really growing. And that's the goal, that's the goal. Here's another, the tiny, maybe the smallest by Genie Attack also. What is this? That's also a 410-E-based SOM. Very small. Very compact. And also, I should also mention very sight, they've been one of our SOM vendors as well. They have a 410-E-based SOM. So is it possible that many of these projects were kind of like kick-started by the 410-C? Communities start doing stuff, and then after they inquire some more advanced stuff, like they contract some designs or something. Of course, when you go, so 410-C really serves, the Dragon Ball really serves a few purposes for us. One is, it's an evaluation vehicle. So if you want to see whether this processor works for you, that's your first, available globally through Aero Electronics. So anybody can go and grab one from Aero. Even a version of it is available through Amazon. So you have a number of ways to get the board. Two is you can, as a maker, as a hobbyist, as an enthusiast, you could use the board to do development, prototyping, a number of things. And three, it's also reference design. So if you're creating a module or if you're creating even a custom board, you can use the Dragon Board as your reference. Because the schematics and everything you need, the documentation, board documentation, everything is available publicly online. So with that, I have to say goodbye because I have to go to my next. But hey, it's been a pleasure having you. Feel free to look around. This is just Snapdragon. We've got a lot more things here. We have a number of things on the, all the modules for Wi-Fi, Bluetooth, LTE. We look around. The wall is a huge wall over there. Wi-Fi, Bluetooth. There's things with our automotive connected car, number of things. Thank you. Thank you for the video.