 So we're here at the Linaro Connect and who are you? I'm David Wrestling and I have the privilege and honor of being the CTO. So what's the latest? The latest, well the latest for me I've been looking at, loosely it's called arm on arm. So arm has mostly been an embedded architecture where you develop software to run on arm on an Intel platform or whatever platform. But through the lifetime of Linaro, which is about eight years now, we've been working on all sorts of piece of software and it's more and more true that that software is developed on arm for arm. So if you look at the server side stuff, the stuff we've done with the developer cloud means that we are building open stack for arm on arm. So I've been looking across all of our activities and thinking about what more do we need to do to enable arm development on arm? So as one example, we're part of Yocto and at the moment when you install the development tools for Yocto, they install on x86 and they run compilers that run on x86 producing Cortex-A code. So we could fix that, kind of. You can go in there, there's a tooling thing I talked with Nico and the folks. So it's really looking around at all the different places where we could be developing code for arm on arm and for other things on arm. And it's part of the next stage of where we're trying to get to. You know, originally, if you think about it, we were just trying to make sure the software ran on arm and ran efficiently on arm. Now we're kind of moving to a phase of it's developed on arm 2. Associated with that, perhaps I should say first, actually it's quite good. An awful lot of software just builds on arm. So all of that work we did early on with the Debian folks, the door of folks, the open Susie folks, means that an awful lot of software is built on arm for arm and we've helped by developer cloud, getting machines out there. One of them, we'd like to see more machines out there being used for this sort of thing. And we've been working with a bunch of folks producing systems. So socio-next, Sincway Subox, 24 cores, it's got PCIe, et cetera. So that's quite a good developer host for that sort of thing. One of the areas just out of interest for me, and I'll describe why, is boot architecture. So a lot of, if you look at the data center, then the system has to boot and be managed in with a particular set of tools, with a particular set of software, because otherwise it can't go in the data center. So that market imposes a boot architecture and a set of boot components. If you look at other areas, it doesn't. You use U-boot here or some other boot over there, and quite how you implement things can vary. That creates what I call sand in the gearbox, right? It slows everything down, and if you've got fragmentation down there, that has worse effects because it stops people using these systems and deploying them and ultimately creating ARM software on them. So one of the efforts that Grant Likely, if you remember Grant, has been working on is an architecture for the embedded space. So it's called the EBR, the Embedded Base Boot Architecture. No, can't be it. Embedded Base Boot Requirements, sorry, we can do that again. So I've been working with Grant, who's been working on something called EBBR, which is the Embedded Base Boot Requirements Document. And this recognises all of the things they learnt from the data center about how that boots, what the interaction between the kernel and the boots software is, where you put environment variables, how you get services from that layer, how that works. But recognising for embedded systems, you use U-boot, for example. So the combination of, and people will continue to do that because they've written all the software and it just works, but there's a set of services that just aren't there in U-boot. For example, the callback services that the kernel might use. There are extensions to U-boot that allow you to run EFI programs to produce that. So really, EBBR is about bringing together those elements to give you a solid boot architecture in the embedded state around U-boot and EFI applications. So there's a process of standardisation, there's a bunch of software that fits there. I don't think that will be implemented everywhere soon. There are way too many devices, but where it's important to get that right is in the gateways, in the edge devices, because these devices are the keepers of security. And in a world where you want to update your software regularly and over the air, the old embedded way of you build a lot of software, you put it in Flash or whatever and then it stays there forever, you need to update it and you need to update it securely, which is the thing that drives the boot architecture. And doing that in the gateway will establish a standard and then it'll migrate out to the devices. Plus, given my background, it's an area I actually understand and I can contribute to and it means I get to play with a piece of code, but I think it's important to establish that architecture and that will enable and allow more development systems as well out there. So the two are kind of linked. So I feel very privileged and lucky to be able to have done videos in the NARA Connect so many years. Yeah, it's been coming quite a while, haven't you? Yeah, but one thing I notice a lot is that maybe still now most of the developers are still using Intel laptops. They are, that's a really good point. So they are starting, though, to it would be really nice to have ARM-based systems, laptops available, right, and folks like me who are not writing code so much, so we don't have a real need to compile. So my main machine is an ARM-based Chromebook that runs Android apps. So my working life is on ARM, but I have to then go to SSH to a little machine to do my development. The SYNQUESA box I talked about, SYNQUESA system, is a nice desk-tied side box-sized box and a number of our engineers have those and are working on things and they've got a decent amount of memory in them for 8 gigabytes, whatever. I even did an interview with Daniel Thompson, he says he was using it exclusively. Yes, yes he is. Now, Daniel's a great proponent of this and the guys inside the NARA and our members are really keen to make this happen. So they've really supported that and getting the point that, say, there's developer cloud and then SYNQUESA boxes, it would be really great if someone built an ARM-based laptop that you could run Linux on because you'd sell, well, how many people are here this week? It's kind of 450, you'd sell 400 of them here because people would just grab them and I think they will come, I think those systems will come. Some of the other guys I've noticed are starting to carry around these ARM-based Windows 10 machines and those have a Linux mode in them such that you can run Linux programs and compile stuff. So you can do some good stuff and the same is happening on Chromebook with Google bringing to that the Linux support, which is awesome. So I think we're at the cusp of the start of enabling ARM developers to develop on ARM. It's a question of the hardware being available and then it will just take off. It's a real inflection point. So one thing that I thought made me very happy at the last Computex because for so many years I've been kind of hoping that ARM would do the marketing in terms of going after the Intel kind of laptop chipsets. But now they're officially saying the Cortex-A76 is a Core i5 class. It's probably marketing now. It sounds like they are going there. I think, well, I can't speak for ARM and ARM's marketing department, but yes, I'm pleased to see that. I think if you look at this year, there's two things that have happened. One is your mobile device has got more and more and more powerful, right? They're very capable devices and also these same companies are producing chips to go in the data center that are really powerful. And so in between there, the real focus of the ARM ecosystem has been on the edge, the mobile devices that we all love and on the cloud. But that development system, the system you carry around, the Chromebook, the whatever has had less priority, but they're coming. And it's sort of generation by generation with some really great stuff. It would be so cool to see ARM take over the whole laptop market. Yeah, I'm a lover of the architects, so architecture. So I want my computing lifestyle to be ARM based. And that may seem a silly thing, but actually it's an incredible thing. If I've been working in this industry a long time and I first met ARM back in the mid-90s, actually. And at that time, the thought that I'd be sitting in a hotel room being interviewed by a chap like you talking about doing all my work on ARM and ARM based laptops is really crazy. But yeah, I think I think it will happen. But in the beginning, ARM was for desktops. It was the original ACON. And then it went down to mobiles. And now it's maybe hopefully coming a little bit back, not just. I mean, yeah, it's 99 percent the mobile, which is amazing. I guess that's true. I guess that's true. Yeah, it started off in the BBC micro and it was for education and the ACON, the ACON RISC machines and all of that. The genius part of the ARM story was their licensing model. The moment they moved to that business model, then that just enabled ARM everywhere. And it arrived just in time for mobile phone technology needing it, right? So that really was the engine that's driven everything. So the evolutionary roots of ARM, the ARM ecosystem, is in mobile. But it has, over the years, spread beyond that. And in fact, the last eight years of Lenaro have been all about moving into new markets, moving into these areas that have traditionally been Intel. So there's been a lot of learning going. I mean, the whole move into the data center has been learning exercise. High performance computing, which is just way sexy, but having conversations with people who are running thousands of processor nodes for months at a time on really difficult modeling of physics problems is amazing. So yeah, anyway, to back to the point, I would love an ARM-based laptop. And we're close. I think it's gonna happen. So now that it's called a Lenaro consumer group, right? Yeah. So does that mean that there would maybe be, that would be part of the consumer group to take care of something more to do with that, if that comes to be a big thing? Well, the consumer group is really built around Android and Google. And in fact, I don't think of them as a segment. I think of each of these things as an ecosystem. So as far as groups within Lenaro, so you have the consumer group, which is all about mobile and Android and that kind of thing. Which used to be mobile group, right? It used to be the mobile group. And then we have the data center group. This is confusing, because we changed the names of things. So it used to be lag, used to be the enterprise group, is now the data center group with an HPC sig. And so that kind of gives you both. We have a light, which is an embedded group, which is focusing on clients such as Zephyr. And then in the middle, we have a group which is based around networking, the edge, the fog called ledge. So we're covering each of these different areas and they're at different stages of maturity. But yeah, it's all interesting stuff as it evolves. And also many, many years now it's been going on the server staff and now has been at the forefront. And hopefully now, the Thunder X2 and all the new Qualcomm chipsets and more, more stuff. Hopefully it isn't going to take over the whole server market. Yeah, you're trying to get me to quote something. There has been year on year, generation on generation. The whole arm in the data center has been maturing. The software for that, all of the work we've done together has really helped move into that market. That's a very difficult market. That's a, you know, to build systems that are deployable into that market, to understand that market is a hard thing. And lots of companies have worked really hard to do it. And the army ecosystem has brought a lot of evolution and diversity into that. But I think it will take a while because corporate buyers, you know, change risk, et cetera. I think what's kind of more interesting is HPC because that's a very different place to be. They understand what they want and they're running very different workloads. So like the Super K, the Post K stuff in Japan is a really interesting example of the arm architecture going places. And they're right here at the Lenaro Connect. Yeah, they are. So what are they looking for out of the Lenaro? They're looking for what everybody looks for from Lenaro, which is, they're looking, it's a place where the arm ecosystem, I know it sounds like it's a bit trite or whatever. It's where the arm ecosystem collaborates, works together on open source software. It's exactly the same point that Jim Davis made today in his keynote and in his interview, which is, look, we have to collaborate. And Lenaro is the place we collaborate. Yeah, which I'm really pleased with that, as you know. I keep describing myself as having the best job in the world. So how would you describe the success of Lenaro so far? And what does the industry say about what you've done? I'm not sure what the industry says. I'm pleased with where Lenaro has got to. Yeah, I mean, and let's be clear about this. It isn't Lenaro, it's its members. It's all of its members. You know, we act as the place where we bring things together and we help that happen. But it's our members adding engineers and doing work and producing processes. I'm really pleased that we are, we kind of are established as the place to collaborate. So we don't have to explain ourselves anymore. I don't have to tell you what Lenaro is. People know Lenaro and that's really good. And our members have been members for a very long time. They've invested a lot in us, which I find slightly humbling because these are big businesses and it's not small amounts of money. This is big investments over time. So yeah, I'm pleased that we're established that we are thought of as the place to collaborate. And I'm really pleased that we're covering all market segments, which for me means I've got a really interesting life because I can be involved in technologies from autonomous driving through machine, I mean, the whole machine learning thing's really interesting. And I say HPCR rather, I think it's good. So from my personal interest level, I'm really happy that we're involved in so many different areas and so many different ecosystems. So what is Lenaro gonna do with the machine learning? Well, look at the announcement announced today. The key, it's the classic problem, is the classic Lenaro problem to be solved is that there's a whole bunch of hardware out there accelerating neural networks differently. There are also a whole bunch of software around neural networks deploying them, inference engines and all of that. And you gotta avoid everybody changing every code base to support every processor or acceleration engine. That is just too much work for any one company. So adopting a framework such that you can plug in support for particular acceleration, being able to exchange formats is exactly that. So the R&N announcement is part of bringing together a framework that works across all acceleration techniques from instructions such as SVE to hardware acceleration like NN2 graphics. So yeah, it's a good real world fragmentation problem that you can solve with better organizing your software and APIs and standards. It seems to be an opportunity with the new 7nm, there's so much space on the die to do all this machinery. There's never enough space on the die, is there? But yeah, yeah, there's a lot going on. But it isn't just at the high sexy end. I mean, the little microcontroller that can recognize your voice saying, okay, Google, right? Those are little cortex M's. As Jim was saying in his keynote and Chris was saying in his keynote, machine learning and neural networks are being applied, will be applied everywhere. It's you're moving from an algorithmic technique of writing software to a data-driven inferencing thing. I mean, yeah, there's all sorts of problems and issues and we have to work across all the communities to transfer the knowledge and do the right thing. But yeah, it's just everywhere. So it will be a big astonishing change to all of our lives. You can think of the medical things, things like looking for diseases, testing, monitoring. No, incredible. So it's very important for the industry to collaborate. Lennaro's the perfect place to do it. Exactly, exactly. And that's what I'm proud of, that Lennaro is the place to collaborate on this sort of thing. And the by-product is I get to work on really interesting technology and meet really interesting technical people. And in the automotive, you're gonna set up a whole thing. Is something's happening there? Yeah, we've been working on certain, on various aspects of automotive vehicles or autonomous vehicles. I mean, there's everything from the sort of processes that are used inside today's car, versus, you know, which is your electric car with advanced driver systems in them, to fully autonomous cars. And obviously the machine learning fits in there because you can't do autonomous vehicles without neural networks and machine learning. So yeah, it all kind of fits together. So hopefully Lennaro, Assignees and employees are excited to join these new groups and do stuff. Yeah, yeah, engineers like new groups and challenges. So yeah, they're very happy.