 what I'm holding here is a high-performance dual socket Thunder X2 arm-based server play. This is going into ODM designs of servers and each one of these have very high performance and it is coherently connected a dual socket system. It has very high memory bandwidth and is our second generation arm-based server in the for the market of cloud and data center. Is each of them up to 56 cores or what was the number? 48? Well this particular one has 32 per core per chip. 32 so those are the big custom cores. Absolutely so this is where high single performance cores. It's also multi-threaded. So is this shipping? It is shipping, yes. So in large quantities? So there are multiple ODMs announced at Computex 2017 with designs ready to ship on this. Already a very large quantity is anybody can build two servers with that right? Yeah. And in there some Thunder X2 servers? Yeah so we have this is another form factor. This is a standard 19 inch in a in a data center rack scenario. So this rack is includes a lot of our second generation Thunder X2 arm-based servers and they are running virtualized mobile infrastructure as well as edge computing service. In this rack it's running all of the virtualization KVM and all of the mobile infrastructure including the virtualized baseband which is virtualizing radio access network also virtualizing the mobile core all of these plus edge services are running on our arm-based server. All right. So what do you talk about here ex-client? Yeah so this is a block diagram showing this edge data center rack. So the rack has servers and also top of rack switch. So the server we have been running the virtualized mobile infrastructure and then the switch here is based on KVM silicon ex-client. So this switch is more than a data center switch because it's also programmable and in this demo is supporting SDN controller and also ONF Linux foundation open network foundation SDN controller and orchestration platform. It is typically used in a spine and leaf fashion so you can scale to very big data center. What are you showing here? Is this a KVM desktop? This is actually not a desktop but this is an edge computing device so it is based on a very scalable arm-based product line that KVM develops. Octane right? Yeah it's Octane TX so Octane TX scales from anywhere between 2 to 24 cores so it supports a very wide variety of system designs. So one example of use case is edge computing. So in edge computing you would put the edge device in outdoor scenarios factories or edge data center so it's a variety of hardware performance level price point and outdoor indoor kind of form factor difference. So with a scalable product line we can address all that. In this particular demo here this edge service is sitting very close to base station so KVM also contributed base station to Telecom Infra project so in this combined... Is there Octane in there? This is Octane Fusion. It's a family where we have a complete base station on a chip design so this is actually a base station which can support more than 100 users. Is that an arm or is it a MIPS? This particular one is MIPS. So this one and edge computing work together so on this arm-based edge computing platform we are running edge services very close to where the mobile data comes through. So for example a user could be streaming a video and the video traffic would be content cached on this edge server so it doesn't have to pass through all that gigabytes of video traffic over the internet. You can serve it at the local area and save the bandwidth and also improve latency. So content caching is running here the mobile core is also running here so this is complete mobile system at the edge using arm-based edge computing. And what is this device right here? So this is another use case of edge device. In this particular case it could be it's an ODM white box it could be SD-WAN for branch office it could be a universal CPE device. What's the CPU in there? So the CPU inside is an arm-based Octane TX. So it is a system on chip using 64-bit ARM. It can be anywhere between 2 to 24 cores. It is doing a bunch of networking. It is basically doing virtualized enterprise networking equipment. So there's a trend of virtualizing the customer premises equipment like firewalls, routers, run optimization, etc. So a service provider can actually offer this as services as opposed to you know their customer buying a separate equipment. What does it say on there? Well this is another demo. So this box once we finish this one this is really you know running a lot of the virtualized services like firewall, router, which pockets enterprise office and this is ARM based. So there's a big ecosystem of ARM based virtualization containers also remote management SDN for service provider to enable this kind of services on ARM. Is it compatible with the future of 5G? Oh absolutely. So all of these demos are related to 5G. So 5G has something like this? Yeah even something like this is related to 5G in the sense that in the sense well this particular one if we use it as an edge computing device as we demoed over there you could have mobile edge computing. So in a 5G scenario you would have the 5G data and then at the at the edge of the network or the radio access part of the network pass the data to an edge computer which could have this form factor depending on the location and deployment scenario to do the local services. So it takes the 5G and turns it into Wi-Fi or something like that? Well yeah so that is a little bit different scenario right so the 5G edge computing scenario one is that for example you could have an edge computing providing security services in a box like this next to the mobile data right next to the base station or radiohead. The other scenario which is the first 5G deployment scenario in US is a fixed wireless scenario. So in a fixed wireless scenario you would have a router at home or office and the broadband interface is not wired it is wireless and it would be through 5G. So you could have actually multiple gigabit of throughput through LTE through 5G to your home router. So that's actually that this demo over here where we have three sets of different offices and each one has a router for their local office but these routers their internet connection is actually through wireless. So with 5G that could be up to multiple gigabit per second bandwidth. Nice so this one is what is a Sprint here? What is they doing? Are they working with you? Yes so we are partnering with Sprint and ARM with Cavium together to work on interesting 5G related project. So in this project we think about in the 5G scenario and also IoT scenario where you you would have to support a lot a large amount of IoT devices. So these devices would pose some scalability requirements because you could need to scale to a lot of connections so your control plane MME portion of the mobile core needs to scale up a lot as compared to the throughput may not be that high. So to support the flexibility of these scalability requirements and also do it very cost effectively we are partnered with Sprint to test a case where we would instantiate EPC as a virtual network function in a public cloud so and also running on containers. So in this case the EPC which is the mobile core runs on a container on ARM based servers that are hosted at a cloud service which is packet.net. So at packet.net they have deployed our ThunderX ARM based server so you can actually rent them for 50 cents per hour and you got a 96 core server and in this project here we instantiating containers as well as EPC on that and if you go to the ARM booth you can actually see a demo by Sprint that they can instantiate EPC on this ARM server in less than two minutes and that's dynamic on you need more capacity you can just spin it up and it's very cost effective. So it's an optimal way of doing virtual servers or SAPI Act? Yeah so the servers are actually you can run virtualize you can run bare metal you can run containers in this particular project we are using containers which is a lot more efficient compared to virtual machines. All right so it's what you call it the container system the ARM server container solution is that what it is? Yeah it's containers running on ARM servers so it is pretty standard but with ARM and with a lot of efficient cores you know we can instantiate containers and in this public cloud environment with a really cost effective way to do additional capacity for even NFV applications. So now as you were saying as the supercomputing the servers and its mass production availability so things are going to more and more stuff is happening. Absolutely. Even in your future configuration where you have a new synergy partner right? Yeah yeah absolutely yeah absolutely yeah right so you know with and then with the high-end ARM based servers we are addressing NFV very cost-effectively with this project where we can instantiate in public cloud and also high performance computing as well as cloud-based applications.