 So we're here in the last day here at the Lenaro Connect 2017 here in San Francisco and how did it go? Actually I think this has been the best connect we've done, for me personally and just generally. I've managed to have a really good mixture of committee meetings, of birds of a feather sessions. I've managed to attend a bunch of technical sessions and I've had some good meetings with members and prospective members. Plus it's a really nice place to come so we've had some good social events and you know I've spent time with Lenaro folks and others. Just friends really, there's a lot of friends here you know. So it's been a great event, I've been here since last Friday, I've been here a week now and I'm flying home tonight. I'm happy to go but it's been a really really good week, very busy week but a good week. A productive week here in the Silicon Valley. Yep, exactly, beautiful and sunny and the hotel they've refurbished and done a good job. Really comfortable place to have the event with the rooms worked and all of that. So now it's seven years? Seven years, I can't believe that. Five years was a bit of a surprise but the last two, what happened to the last two childbacks? How come it happened so fast? Suddenly it's seven. I think it was Trump, the campaign. Don't get me to say anything political childbacks, I'm not doing that. Now we shan't talk about it, let's keep this nice and friendly. But did it a little bit start with the there was some attention around the mobile stuff and then all these other things. There's some new ideas, you have some new ideas, you talk about automotive. So in terms of the broad sweep of things, mobile is important but maturing and actually the work with Google and Treble and you saw the keynotes, you just see where we're going there and that's great. That's really good, the stuff that's happening there. Google really wants to have Linaro help with the whole Treble stuff, right? Yeah, they recognise the part we can play. Obviously there's the wider partnership and the guys like Corcom and Hikey and the rest of it. And that's mobile, that's very relevant to mobile. But in other areas, so enterprise, having a day centre day was really good. And Martin did a great job pulling it all together and it's showing where we've got to, you know, from us. Well, the enterprise group was the first group we formed and that really changed Linaro and adding more groups has really broadened what Linaro does. But for enterprise, there's a lot of, you know, all that work is coming together and you're seeing some great systems and so the day centre day was really, really good. The other thread that's been running through, so obviously it's the CTO, I like new stuff, old stuff done, you know, let's move on. And a couple of new things that are really good. One of them is HPC, so the work that Canta Vicaria, who's on my, I only have a little team, Canta's one of them. She's on your team with the strategy? Yes, my office, okay, right, so my office, I have a little office of the CTO. It sounds grand, but there's three of them. There's Canta Vicaria, there's Rob Herring, who's more on the kernel, Android side, and there's Nicholas Petra, who's been working to make Linux really small. He's trying to get it to run in the smallest memory, out of flash, smallest memory footprint, flat memory map. How small would he get it so far? Oh dear, I have to cut. Well, he's a little shy, that's okay. He's very quickly going from one thing to the other. Yeah, but I can't quite remember. He's using, you should look at his talk on video, he's using an STM Cortex-M part. Running Linux in the Cortex-M. Yeah, and he's compiling it fdpix, so basically it's a flat memory map. You can have loadable images. So the kernel is running out of flash, and then it's booting in a small amount of memory. I think that part has 640k, but it's so he's using external memory. I think he's about 800k running a very simple environment, simple script. So you can basically boot it and do ls, and cat slash procs, like CPU and stuff. That could be a very revolutionary for some kind of thing. Yeah, well, where he's coming from is that Linux is very rich. It's rich in device drivers, it's rich in applications. And so you can enable it down to a pretty small footprint. The problem he's facing is that persuading the kernel developers that making changes that don't actually benefit them, but benefit the real embedded guys are worthwhile. But he's doing well. A lot of his stuff is being merged in the next merge window, or this merge window. So that's good. So good stuff. And you should watch his video. He explains what he's doing. Does he have strong opinions about the strategy? Well, he cares deeply. He's on your team. Yeah, yeah. This whole IoT-L, IoT Linux, he's been pursuing, he's been working at it and persuading people. Is it an alternative to Zephyr, or is it useful for Zephyr? Well, it's useful because it keeps Zephyr in its place. So Zephyr is a very small RTOS, and it shouldn't get too big. It shouldn't add everything into it. It shouldn't try to be everything. So it's nice to have a little crossover. And then you know it's very active in Zephyr. Yeah, yeah. The idea behind Zephyr is that it's kind of like the Linux kernel of the embedded world. And the key thing about it isn't the code base, it's actually the maintainership and pulling all that together. And the code base is coming along, and it's the kind of where I thought it would be after about a year's effort. Companies are starting to, well, they're supporting it, and they're starting to talk about releasing versions to support their process. So Zephyr's going well, that's the light group. It's been used a lot in this new open source foundries. Yeah, because it's a useful IoT client. It runs on a tiny little Bluetooth part of things. All you need is Bluetooth and a little bits and pieces, and then it can run a bit more. So I think it's worked out really well, that one. And so the open source foundries is going to be an exciting thing? Yeah, it's exciting. It's an opportunity. When we started Lunaro, I hope there's a couple of, besides the collaboration thing, I was hoping that there would be some business opportunities, and that's one part of it. So this is kind of where that came from. It's an opportunity, and the thing is that Lunaro's mission is about this collaboration. So it doesn't quite fit in Lunaro, because we don't want to pollute that mission as it were. We want to work together, and once you get products in there, it can get a little sticky. But what they've done has been really great. So spinning that out as a separate business opportunity makes sense. The other area I wanted to encourage, and you've seen a little bit of that here, is the relationship with universities and academia and the maker community. So the 96 boards open hours. I didn't really get to see much of it, but there was a room full of students. It's packed and packed. Packed and packed. And the professors are giving them, you know, they cancel lectures to let them be here. You know, I'm as passionate as Mad Dog about the young graduates and students, because they're the entrepreneurs of tomorrow. They're the ones who are going to do things I can't imagine. So I was really good to see that. We'll probably do more around that. But anyway, with Lunaro, I wanted that relationship. I wanted the business opportunities and the rest of it. And that's kind of where associate membership is coming in as well. We want to engage with the wider community. If you look at the history of Lunaro, it's been about the arm ecosystem getting together and solving problems together. Now it's becoming much more about involving everyone in that arm ecosystem. The wider, you know, the wider ecosystem, the wider members of us. It's about widening that participation. So with the associate system, there might be dozens or hundreds more, kind of like members. We've been taking it pretty steadily. You know, we've not assumed there'll be thousands or hundreds or tens, but we've talked to a lot of companies we're interested, universities are interested, and we'll build it up slowly because you're kind of building around something and you want to be able to scale it. But I think it's an interesting and exciting opportunity. And maybe next you'll have Audi and Fiat Ford, Audi's coming out with that. So as I say, I'm the CTO, I like new things. I've spent the last year or so talking with a bunch of folks in automotive. I'm convinced that we could do something to help. So the thing about arm in automotive, it's already used a lot, but arm works well in ecologies where there are standards, where, you know, and we think there's a bunch of stuff that we could be joining in with. We, you know, you don't always invent, we don't always invent standards. You follow them and stuff. So we've had a big automotive presence here this week and a lot of discussions around automotive, including in the technical steering committee. And basically we're going to be doing more. So it's one of those watch this space. Because what is a free, free software foundation doing? And what is linear are going to do in this space? Like, what is your role? The foundation has the automotive grade Linux. In fact, Dan Couchy was here from there and we had a couple of conversations. And what they're doing is great, you know, they're they're stuff around Adas and, and, you know, infotainment, really good. And, and they're, you know, it's actually in cars, I believe. So that's great. There's a wider thing of what does the whole system look like. But, but, and the issue is one of the issues solve is the fact that vehicles and not just cars, ships, boats, planes, drones, you know, that you name it, trucks, they they there's a whole bunch of functionality they're going to need. And the only way to get that functionality is to include open source, right? There's Matt, it's there, it's in open source. It's happening in open source. It's part of the drift of technology from the data center, really. And once you add in the need for over the air updates and, you know, the cars are mobile data center or a robot with wheels or whatever. So the issue is really how you design a system to mix high criticality systems like your braking systems, your safety systems with open source systems that are doing some other stuff, you know. So it's a very interesting place to research and we're really mapping the territory at the moment. We're looking around at what's happening and thinking what layers and APIs and systems and technologies around and how it goes. So, yeah, because self-driving cars use some kind of computer vision and there's AI. Well, OK, if you if you think about it, you need to have 360 degree vision, LiDAR, low end radar, infrared, you know, and you need to integrate that view with a bunch of machine learning that recognizes objects, tracks them and does stuff. Now, the early parts of this are kind of like keeping in the lane, accident avoidance, but so there's a bunch of what are now enterprise class software that needs to run on the car. You can't afford to run all that in the cloud because, you know, your car is driving along and the connection goes down for, you know, a few hundred yards and the car can't make a decision. So there's a lot of local decision making going on. So it's a very, very interesting area of computing. So in AI, in general, the NARO is going to be involved because it was one of the key notes that said you can't just have China compete with the US or with some other. I'm not quite sure. It used to be an open source AI. Otherwise, we would be in trouble. My DNA is open source. I'm going to say, you know, that's where the research is going. Yes, there's a bunch of stuff that we're looking at around AI. AI is a very loose term. It's quite fuzzy and meaningless. Is new areas a new social system? Oh, yeah. I'm sure that the ARM ecosystem is thinking about it. If you think about it, one of the key attributes of the ARM ecosystem is hardware acceleration. The reason why the ARM ecosystem competes so well in ecosystems is you get parallel evolution. You haven't just got one way of doing things. Companies are all trying different things. They're trying intelligent fabric. They're trying this accelerator. And, you know, AI is not just, it's not about GPUs. I won't be about GPUs, but it's about deep, deep pipelines of processing, you know, doing transformations across deep pipes of data. Going back to car vision, there's a bunch of processing near the sensor, right? It's like the human brains. Human eyeballs do a bunch of processing. And then your main brain integrates the results. Well, think of a car architecture that kind of does that. But then think of security and think of the safety side of that. You know, it's interesting. I've had a lot of very interesting conversations. Cool. So looking forward to the next video and the next linear. Yeah, well, Charback, you'll be there. It's in six months. It's in Hong Kong. Right now I need to recover. I need to go home, walk the dog, relax. But I've had a really good time. I'm glad to be flying home. I love California, but I'm glad to be going home and tired, but happy. Are you going to post the dog on Twitter on Google Plus or? I do on Google Plus. I tend to post amongst friends. The dog is a Dalmatian called Sid, or Sydney, Sid or Sydney. So, yeah, I post there. In fact, you should see him. You see him on yours. We're friends. OK.