 So we're here with John Masters and this is the Enterprise Edition 96 boards with an AMD ARM processor. So what's going on? That's right. So what I'm holding here is the first, hopefully of many, 96 board of the Enterprise flavor. So there's two different kinds of 96 board. There's the Consumer Edition and then there's the Enterprise Edition. And the purpose of the Enterprise Edition is to provide a slightly larger, both I guess in physical form factor, but also in functionality, kind of a more enterprise server-focused design. And what that means is a combination of server grade hardware. So what you have here on this board is an AMD Seattle A1100 SoC, which is an 8 core ARM Cortex A57 processor. And then you've got the ability to put your own memory on here, whatever configuration you want. You've got a bunch of SATA ports, got a couple of gigi-network interfaces, you've got USB, you've got PCI. You've got all the kinds of things that you would want on a server platform. And in many ways, from a software perspective, it looks like the much more expensive server reference platforms that I've been working with, with AMD on for the last few years. But the interesting thing here is that this is going to be a more affordable platform. I don't think they've given precise pricing yet, but you're looking at something that's affordable for an end developer and affordable certainly for companies. You're not paying thousands of dollars for this board, far from it. And you're going to have something that is physically this size. So you're going to have... There was a prototype here at the Nenar Connect. Somebody was showing off the prototype, right? Yeah, they had a mechanical prototype. And then the actual boards, the kind of real production boards are just a few weeks to maybe a month or two away from starting to surface. I'm hoping to have some at the ARM TechCon in November and be able to show a little bit more in what we're planning to do there. The interesting thing from my perspective is that I got involved fairly early on with the design for the specification for the 96 boards and a price series. So there's a specification. It's an open specification that anyone can take and use. The beautiful thing is anyone can make one of these boards. This one happens to have an AMD processor. Somebody could make a different one if they want. It's an open specification, but it's been engineered from the beginning. And I got involved fairly early on in that in the first few days of the specification last year. And made sure ahead of time that it would run a lot of the enterprise software that you see out there. So when you get it, you can just take the Red Hat Enterprise Linux server for ARM development preview that we've put out. You can put it on a USB stick and you can do a live install and off you go. You can put a DVD drive in this and do an install. So all the kinds of things that you would expect from a real kind of server you would have today, you can just take this, plug in your media and off you go. We've got USB stick. Correct. Live install. Boom, off you go. Boom, and it installs, it just works. That's it. That's it. And we've been doing those tasks since last year on the various, if you like, prototype platforms that fed into this design. So we've we've tested extensively doing USB, you can do a live USB if you wanted to. So actually all the kinds of use cases that people have today, like on your laptop, you want to try a new Linux release, you can get a live USB stick, put it in and boot off it. All that stuff works on some of these emerging ARM servers. That's going to be really cool. What's part of the spec that you think is really important? Like all the connectors are the same places or what else is? Yeah, so so the standard form factor is very important that it can fit into the kind of cases that people are used to getting. Fits in a standard case of. It can fit in the standard. I don't think this is mini ITX is one of the slightly smaller, not Nano ITX, one of those kind of ITX series specifications. It will fit into those cases. Things that interest me about it is that it's got PCI, which, you know, when I started the ARM server in the ARM server racket, I guess, four or five years ago and started playing around with this stuff, people said, well, ARM is going to be all about SOCs, people integrating IP devices. Why would you ever build a PCI? Why would you have all these different external devices? That's not what ARM servers are going to be about. But I think a lot of people kind of jump ahead of the industry. What the industry wants to begin with or thinks it wants, more importantly, it's what the industry thinks it wants, are high-end servers that look a lot like what they know, where they can just take a PCI card that they already have. One example we have, we work quite closely with Melanox. They have some very nice 10 gigabit ethernet hardware. And so if you don't want to use the gigabit interfaces on here, you can get a PCI card and put it in. One thing I think is going to be very interesting from a developer point of view is putting in the AMD Radeon PCI graphics card, which we've tested. And the cool thing is, even though this is not a desktop, it's not intended to be a desktop platform for around, for the range of $500 to $1,000, not more, in that kind of window, you're going to be able to get one of these in a case, put a nice graphics card in there, some decent memory, decent disk, all this stuff, kind of money you might spend on a desktop anyway. And you're going to be able to run a Linux distribution of your choice on this system as a developer with a graphical desktop, with all these features. And I think that's going to be very interesting. So it won't just be people doing servers, but it will also be people who want to have a traditional PC style system that they can build with this now. That's very cool. So this is the ARM desktop. This is 64-bit. Yeah. Up the core. Eight core. You could certainly build a desktop. You could certainly put any PCI GPU on here. And in fact, the desktop team in Red Hat have started testing out different. It's kind of a fun thing, right? We do servers. We don't really do ARM desktops. I don't think that's the primary focus for this. But a lot of people here at Connect have been saying, gee, if I can buy this for hundreds of dollars, not thousands of dollars, and if I can put any graphics card I want in, if I can run my GNOME desktop or my KDE desktop or XFCE or whatever desktop I want as a Linux person, and if I can just run the same applications that I'm used to running, now I could be a developer at home or I could be a Linux community developer who's not an ARM-focused developer. I could be Linus Torvalds. I could be, well, I wish, right? But let's say that someone like Linus could say he wants to try building and testing ARM, builds the kernel more easily. He could even have one of these on his desk and run it as a desktop if he wanted to. And that's a cool thing. Once you can make this hardware available, both for people who want to build interesting servers and have a reference platform, and then also build their own desktop. And one more thing I would say about this platform that's particularly interesting is everything is open. Specifications open. No chip vendor is going to publish the schematics for how to build their processor. So some of the hardware is kind of standard for the industry, but the firmware is open. All of the reference software is open. So what you can do as a software or even firmware developer is you can download everything that goes into the system. And you can hack on it and you can understand it. What we're going to do in Lunar that I think is really cool is we're going to work on new technologies. We're going to provide reference software and reference firmware that runs on boards like this. And this is the first one. And I love AMD, and they're great. But there are many other companies out there building interesting ARM devices. And there will be many, many more like this over the next year or two. So I've been hearing, and it sounds like, and it looks like that the ARM server is going to explode. I think so. I think that the thing with ARM servers is you have to make the hardware available. The biggest issue we've had is we've got some great operating system support out there now. We've got some great applications. I've been demoing at this event Apache Spark. OK, doing real-time Twitter streaming analytics. So it just works. You can take it upstream, Apache Spark release. It's the next thing after Hadoop. Everyone's kind of really into Apache Spark now. There'll probably be something even more shiny next year. But you can take it. It's a Java app. It just works. It works really well. You can build a cluster. I've done all these things, but that's because I've got hardware. And the biggest problem is getting hardware to developers. So now, if you have inexpensive reference platforms that show people how to do it the right way, and they can get those, and they can port their software, and they can do the development, and they can test things out, that's going to help with the ecosystem. And then as the ecosystem builds, you're going to see more mainstream hardware vendors selling systems with trips like this one in there. So we're very close to all the linear engineers being able to have ARM-powered desktops and laptops, maybe. And who knows, maybe Nalibaba or Amazon or somebody just starts building whole server parks, just ARM-powered. So I saw a really interesting demo here this week. I saw the guys from the CentOS community were here actually showing Zen and Cent running on ARM. We also have great support for KVM. We have Red Hat's working on its own commercial stuff. But just as a community example, what they had there was Cent running on Zen. Now, a lot of these public clouds that you mentioned there, they run on a mixture of hypervisors, both Zen and KVM. Great support there. OpenStack's running great on these things. So if I am one of those public clouds, there's really nothing to stop me from saying, let me go and deploy something, except availability of hardware. So as we see more systems like this, it'll help with development. And then it will help with the definition of open compute compatible server platforms and mainstream Tier 1 OEMs, building very large systems that these cloud vendors can roll out. Because they are planning out to build a bunch of server parks. And they have to consider very, very strongly to jump on the ARM. On the ARM, yeah. And is it ready? It's ready. It's ready. I think if you wanted to build, I think it would be aggressive now, but you're not too early to the party. Two years time, you might be late to the party. Now is a really good time to go and build a 10,000 node cluster. If you're someone like one of those public cloud companies, 10,000 systems is a small number. So go build a room full of racks, work with one of these vendors, build, work with software partners. You don't have to work with me. I'd love to work with you. But work with some of us in the industry and go and build some of these systems. And let's see where we can go. And let's see how much money we can save. That's absolutely true. Absolutely true. And I look forward to the opportunity to build some of these things.