 So we're here at the demo Friday, so did you succeed if you make the desktop? Oh yes, we made some very nice demos and the desktop is one of them. As you can see here, it looks like a normal desktop box and it acts like a normal desktop box. And that's the one with the Quark Core A72. You can even play some 3D games on it if you want to. So all this open graphics drivers, open firmware, not just a piece of proprietary... 100% open? Yes. There's nothing closed? Nothing closed and... It's a Radeon GPU? We have several GPUs running. Originally it was a Radeon but that broke down during testing, probably overheating. Right now it has an NVIDIA GPU with the new VOTE driver. Yes, that's what it's running, so... Not as many as a Radeon? Yes, the Radeon open drivers are in a much better shape right now. So this is a Macchiato bin, a Marvell, what do you think about this performance? Are you getting stuff? Can you be a building only or a code with this? The performance is pretty good. I managed to build a kernel with pretty much all config options enabled as modules in less than 10 minutes. Which is not great but definitely quite good for an ARM box. How long does it take with the Intel thing? How long does it take with the Intel thing? It takes around 3 minutes. 3 minutes? I know you have 10 minutes. Yes, it still has a bit of catching up to do but this is perfectly usable. Can you show the windows? What are you doing here and there and there? Yes, here essentially I'm just showing what it is. So it's an ARC64 box and it's running mainline kernel 4.14. Here it's just showing the processor usage. So CPUs are not exactly overloaded. There's also still lots of memory available. Here's just a little 3D game for demonstration. What's the type of Linux you have running on there? That's OpenMantreva, my personal favorite. You always use that? Whenever I can. There's a couple of specific use cases where other things are too better. If I want to go to an extremely low-end device I'd probably start with OpenMantre. If I wanted to build a phone I'd probably use Android because of the apps. But for most things I really like OpenMantreva and one of the nice things about it is that it's a very flexible team so we could, for example, move the default compiler to Clang without having to spend months or years discussing it in committees or something. Let's check your ARM part laptop. Yeah, so that one you have too right here? That is essentially running the same stuff. It's not yet quite on 4.14, but kind of 4.11 is also not that old. It's also ready to run some light 3D stuff. So this box is... Yeah, let's open it up. So you can see it's originally a pie top but I've replaced the inside because the pie is just too slow for a regular desktop. So instead of the pie we are powering it with the DragonBot 820 and that required a couple of other changes. We also built in another SSD so we have a couple of terabytes of storage. Two terabytes? Yeah. And what's over there on the left? That's just the output from launching the game. So nothing overly interesting. I mean here. Ah, that? That's actually an interesting part that comes with the pie top. So this essentially takes the HDMI output off the board and input events from the keyboard and touchpad and translates them into signals the hardware here can understand. So the screen is not an HDMI display but it's powered by the HDMI port and the keyboard and touchpad just drive us regular USB devices. So your laptop is pretty big. Can you fit the Macata bin in there? That would be fun. Can you fit the KVM Thunder X2 or the new Coltom server? That's a good one. That's a really good one. I'm really looking forward to getting hold of it but chances are I'll never get hold of it. Those things are just too expensive. Can you build a cooling system that would fit in your Predator laptop? I'm pretty sure because there's an i7 in there and that gets hotter than even the biggest. That's the system you use for cooking at home. Yes, right. So when I want to cook, I just put a pen on top of the laptop and then I run and make this J. All right. And so it will happen. You have it right here. You solve it. You have the laptop. You have the desktop. And this is the Leonardo Connect 2017. It's a historic day, right? A little bit. Yes. So the desktop is essentially ready because it doesn't have any special parts that need to be built for everything. So if someone told me I need one right now, I could build one within a day. For the laptop, you need to figure out how we can make it more easily reproducible. We've been talking about this in videos for the last seven years. Yes. I've always been saying someone needs to do it. Someone needs to do it. All right. But what more do you need? Do you want more SoCs? Faster SoCs? Yes. It would definitely be nice to have something even faster than this. And this is good enough for most users, but if you really want to heavily build stuff, you probably want an even faster SoC. It would be nice. It's over there. It's a 48 core. Let's jump over here. Yeah, sure. So there's Thunder X. I mean, Thunder X 2. Somewhere around here is a new fast console. Yeah. It's really cool. It's like into a right here. Hardly the very easy. That's a powerful 48 core. It would be nice to have that at an affordable price. So it could be built into a regular desktop. It's a nice big chip. Right now, I think the price is pretty much the blocker for this. With one of those, you can probably build a Linux in one minute. Probably even less. And that would be three times faster than the i7. I wouldn't be surprised. Cool. So that's for the next neural connect, right? That would be fun, but realistically, I don't get access to this type of board. And the guys to do are very much focused on the server space, so I don't see them playing around, building a desktop machine out of it. And over here you have something that says AUSP TV. Yes. Another demo, but I can't show that right now because we're having bandwidth problems. But the idea there is that we take a TV input signal from a satellite dish or even a terrestrial antenna in Switzerland, stream it over the network and play it on AUSP TV right here or wherever we want to be. So what have you done with AUSP TV? You've been working on this a little bit? Yes. So right now what has already been done is a little disconnected from AUSP TV. It's essentially implementing all the stuff to get signals from DVB, making sure that it's available in code that is licensed and evaded, AOSP can accept. The next step there will be integrating it into the AUSP TV interface. So it can be done right from the AUSP interface with remote controls instead of typing in commands to get the stream and to play it back. So basically it's an open source version of Android TV? Yes, essentially that's what it is. How much of the Android TV are you able to get into the AUSP TV? It depends on how closely you look at licensing. There's a couple of things you can do to get the full Android TV experience on AOSP which essentially involves getting the binary blobs for the Nexus player and copying them around. But obviously those things aren't licensed very nicely so there's more stuff that needs to be... How can you take the Intel binary blobs and just make them work on ARM things? The thing is they aren't binary blobs. Most of the stuff in Android TV is implemented in Java while I'm not the biggest fan of Java. For most things one of the nice things about it is you can just take code from an X86 box and run it on an ARM box or vice versa. And you also have a special phone. You don't have it right here, no? Yes, that's in my hotel room. You work in all kinds of different things? Yes, sure. It has to do with Android. I'm very much a believer in something that needs to be done. If nobody else is doing it, I'll do it. You do it. So we want the desktop. You do it. But it needs to be smoothened out and let's get it out on the market, right? People need to be able to buy this. Right. And the laptop... Working on that too. Let's get the laptop out so people can buy it, hopefully. And choose any SoC they want, just plug it in. That's the 96 boards idea, right? Yes, that would be really nice. At the moment with the laptop, it's not possible because we had to use a different form factor because the particular SoC we wanted in order to get open graphics drivers wasn't available in the right form factor. But that's probably changing in the future. And Android has this project treble thing going on. Yes, that is really interesting. Maybe thanks to you and thanks to the Naro, maybe every phone and every Android TV box, every Android, everything that gets Android 8 will be updated forever automatically. That would be nice. I don't really see that happening, but probably it does mean that updates will come a lot faster than before. And the Naro has a role to play. You have a role to play. You're working with that stuff, right? Yes. What are you doing with that stuff? Most of the things that Naro is doing for it at the moment are related to kernel maintenance. One of the things in Treble is standardizing on the LTS kernels to make sure that security updates for the kernels can be delivered to all devices. Those kernels are, for the most part, maintained inside Naro. We are also starting to do some user space work to support reference implementations of certain updates and things. But for the most part, our involvement is on the kernel side. So your role is not only in the mobile group. You're also doing stuff in the home group because you're doing the AOSP TV. Yes. There is a lot more things to happen in the mobile in the future even though Android is already the big king, right? It needs to be... Yeah, of course. There needs to be a lot of things to happen in the mobile group. If Android just stalls, it will probably remain the best thing there is on the mobile for another year or two. But if people stop developing it, if people stop innovating, something else will come up and take over even if I have to do it. We were talking a bunch of years ago, three years ago, we were talking about maybe running Debian or other kind of apps natively on Android. Maybe even we were talking about crazy stuff like running Windows on Android. Maybe with a Snapdragon 845, they will be able to dual boot or some kind of thing. And maybe you even run Windows on top of Android. Maybe you run iOS apps on top of Android. What about all these crazy ideas? Yes, so I still think some of that is possible. Like getting normal Linux applications to run isn't all that complicated. What's missing is essentially integration between the different graphics stacks. Android uses surface finger while regular Linux for the most part uses X11. Some distributions are switching to Valen, but neither of those is compatible and you can just connect it into a surface finger. But they're not all that different. They're all based on OpenGL these days. So there's a good chance that at some point someone will come up with a proper solution there. And as for the other operating systems, I don't really see why you would want to do it, but if you look at the VineGit repositories, for example, there's already code to support Android. So if you had some Windows applications that were built for the right processor, which is currently rare, you could run them. And in terms of iOS, there's GNU step which implements a lot of the APIs and Swift is open source these days. So if people just implemented a couple of wrappers in between, there's a good chance it could work. If you have time, should we go to the Steve Jobs Theatre? We can go to the Apple headquarters and convince them to support open source? They would do no such thing. We can convince them. I'm a programmer, not a politician. Maybe you should run for Congress, or run for president or something. Don't you think you will have a chance? No way. I mean, even if I wanted to, which I don't, there's no way people would vote for someone who looks like me, for someone who talks like me. I think you have the real experience, everyday experience on the ground. They are living in another planet. Maybe you can make an app or software that decides for people what needs to be the politics of the future. Yes, right. I call it brainwashing 1.0. I hear Apple is really interested in putting that on their phones. They probably already have it. They just don't admit to it. They have control on a lot of Apple fanboy brains, that's for sure. How do we convert them? We need to... It's a good question. I still don't really understand what Apple's appeal is. I mean, their devices look good, but they don't hold up to comparison with other devices, if you look at the technology. I did a really cool video with the purism. They have this deviant phone coming out. They want to get the deviant phone to work. Do you think there's a good chance it's going to happen? It's going to work? That it's certainly going to work, but I think it will be hard to convince people to start using it because there's just too many apps that people have come to expect. They just need to have 3,000 customers and they can get it funded. It's just 3,000 guys that want to have a deviant phone. Then this is lost. Who knows, maybe it gets converted into 3 million users. That would be awesome. Yeah, I think we need some more diversity in mobile space. Would you like to send a message to all the industry and all the chip makers that they should all join the NARO? Yeah, of course. Everyone should join the NARO if you can afford it, of course. And not just that, even regardless of whether or not you're joining the NARO, please do open source drivers. They make our life much easier. And please don't ignore the desktop and laptop market the way it's always been done. You can see over there that it's possible to do something, so let's do it. Which property drivers do you really want to get open source right now? Mali is probably the biggest pain point because that's in most SRCs It's not even at a point where you can make a display work using pure open source stuff. Can you speak with a British accent and send a message to the friends at ARM? Hello. They already know where I stand on this and a lot of people inside ARM agree, but unfortunately their decision makers aren't there yet. I think it will be a surprise, that Christmas surprise this year will be an open source model. That would be the best Christmas gift to me ever.