Gadgetory

everyone I'm Steve from gamers Nexus Don that we're at GDC 2016 at Nvidia's booth and now we're looking at g-force now which is a game streaming service I'm joined by Chad Cooper from Nvidia to talk about that he's a build engineer working on GeForce now before we get to this coverage all of our GDC 2016 coverage is brought to you by raw fury is it gone or platformer game what you see on the screen now Chad so what's the top level overview of GeForce now we know there was this grid thing before is this grid just with a new name or what's going on and this is a latest version of it right now it's available on NVIDIA shield branded devices it's currently $7.99 a month you get access to a huge library of games and then we also sell titles as well and what's unique about that is if you buy the title you can play it immediately so you don't have to worry about downloading it or driver updates or patches or anything like that you launch it it works and then on the back end we also email you the license key to the game so you can download on your PC if you want so you get both options there what about so what about input latency things like that with with regard to when I'm playing the game I'm streaming it I push a button on the controller how long does it take to respond on the screen input latency we try to keep it down to be on console level and we've that's been a big challenge at Nvidia and what our engineers have done is they put a hardware encoder on the actual GPU so as soon as those images are loaded up into the frame buffer they're then compressed and encoded immediately on the GPU in about 2 to 5 milliseconds depending on the resolution and the frame rate that we're going to be sending through it and then on the client side we have that same encoding technology there and it's decoding it real quick so that's what makes it extremely snappy so that's that's kind of how we solved it and how about a millisecond target for input latency specifically target wise we try to keep it on par with consoles just really a low latency target as fast as possible we deploy data centers around the globe to also assist with this and we do work with ISPs to ensure that people get the latency less experience and then server-side so there's the as they used to be called grid but I guess the g4 is now servers are those set up in different locations around the country or how are you setting those up correct right now we have them a few players around the US and we also have them around the world as well so in data centers like Singapore in Tokyo we also have them there and in Europe in Ireland so it's it's a global right now and it's available does the server basically do all the graphics processing all the computing and then just ship a package sort of video to the computer to the client is that how it works that's exactly how it works so what we do is we put our high-end Tesla great GPUs in the cloud and then we create a virtual machine that's dedicated to that end user they've got dedicated CPU cores RAM even provision I ops so it ensures that they have the best gaming experience possible from the cloud and then other than that we only support shill brand devices because of our decoder so we cover it from end to end to ensure that you get the best experience possible trade shows as as we know maybe the viewers don't necessarily know but trade shows have historically terrible internet it's really hard to talk outside the show what do you use here to work with GeForce now and you're demoing yeah for this event what we do is we piggyback off some of the technology that we've been working on on the engineering side so a lot of our engineers and architects have idea is to deploy single rack units and small data centers so it's a proof-of-concept so we've done is we've taken that concept and we shrunk it down into essentially like a monster gaming PC and so that's what we brought to the booth so we have a whole virtual environment or a whole data center tons of VMs running all the networking all the beyonds going all at once and it's streaming to these end devices now for our game saves and our entitlement and all that stuff it piggy backs off of AWS still so it goes back to our datacenters that release out for AWS just for that and then those user saves come down to this this local server and then you're able to play your game where you left off even if you logged in with your own account here so it's kind of neat that is cool what's the GPU set up inside the GPU that we have in this config is our M 5000 so basically a GM 204 is Maxwell class so it's equivalent to a gtx 980 effectively so that's that's what you're getting in this demo to the rig because it's pretty powerful yeah yeah that's pretty powerful that you mentioned bison staying with AWS with for when you communicate to AWS that's one other question I have so most these games now obviously have some sort of DRM or licensing how do you guys work with that do you do you talk to the publishers or what's the deal there yeah we have a great relationship with the publishers so we've had that for years working with them and so each publisher we handle it differently and so it's really kind of up to them we are releasing a grid SDK or a cloud gaming gets SDK for the developers to use for onboarding games to the cloud to make things easier and we can provide I'll provide you more information I'll send you a link so cool so more information linked in the description below if you're curious we've got an article on this and then of course there's some pretty big game works and news that came out yesterday we've got a full 2000 plus weird article detailing all that if you're curious how it works especially with things like voxel accelerated ambient occlusion and all that so links to the description below thank you for watching we'll see you all next time