WEBVTT 00:00:00.800 --> 00:00:09.360 AMD brought us to the suite here at CES 2017 to prove once and for all that 00:00:07.359 --> 00:00:12.800 Ryzen is real. It is physical. It is tangible. You can touch it. Well, you 00:00:10.960 --> 00:00:16.880 can't touch it. But but but I could touch it because I'm here if I as long 00:00:14.960 --> 00:00:21.039 as I didn't care about like EST shocking it cuz these floors are carpeted. Not 00:00:18.880 --> 00:00:24.800 only do they have the couple of systems that I'm going to show you here, but 00:00:22.400 --> 00:00:31.760 they actually have systems from over a dozen of their partners. everyone from 00:00:28.560 --> 00:00:33.600 PC case gear to 00:00:31.760 --> 00:00:37.440 what the devil is this thing? Whatever it is, it's water cooled. It's red. It 00:00:35.280 --> 00:00:41.200 looks actually pretty freaking bananas. And the logo is in Chinese, so I'm not 00:00:39.840 --> 00:00:47.920 going to try too hard to pronounce that one. Anyway, let's get let's get back over here to the demos here. The systems 00:00:45.120 --> 00:00:52.559 in front of me right now are the ones from the AMD live stream about a month 00:00:50.399 --> 00:00:58.239 ago. Not they they're not rebuilt the same spec. They are the exact systems 00:00:55.760 --> 00:01:02.800 because that demo raised a lot of questions. Did AMD change this? Did they 00:01:00.719 --> 00:01:09.760 change that? What clock speeds is it running at? So, we have the answers to a 00:01:06.159 --> 00:01:12.799 significant chunk of that speculation 00:01:09.760 --> 00:01:15.200 right now. So, this one right here is 00:01:12.799 --> 00:01:21.759 running an 8 core 16thread. They let me take the panels off. 8 core 16thread 00:01:18.400 --> 00:01:23.280 Ryzen CPU on a validation level 00:01:21.759 --> 00:01:27.280 motherboard. So, this is like a reference design. You can't actually buy 00:01:25.119 --> 00:01:31.280 this board. They've got dual channel DDR4 memory in here. They're running a 00:01:29.200 --> 00:01:36.560 Titan XP because AMD are a bunch of smart folks. They know that RX480 is not 00:01:33.920 --> 00:01:40.799 going to prove that this CPU can run with the biggest, fastest GPU on the 00:01:38.720 --> 00:01:48.159 market today. They had to put that in there. It is what it is. But the point 00:01:44.240 --> 00:01:50.320 is, it's the real deal. Now, we have a 00:01:48.159 --> 00:01:57.600 look at this guy. AMD claimed they had a sixtyn00 K, a 00:01:54.560 --> 00:01:59.759 thousand processor from Intel running on 00:01:57.600 --> 00:02:03.920 legit actual production hardware. Oh, look, that's what it is. So, the demo 00:02:02.240 --> 00:02:10.560 that we're looking at right now is Battlefield 1 running on a six-t with a 00:02:07.759 --> 00:02:13.360 Titan XP. Otherwise, the same hardware. Yes, the case is different. That does 00:02:12.000 --> 00:02:19.520 not really affect the performance very much. If anything, that's an advantage for the Intel system because thanks to 00:02:17.200 --> 00:02:24.480 GPU boost, that card will run a little faster if it's cooled better. And right 00:02:22.400 --> 00:02:28.400 here we have it. Frame rate counter. Battlefield 1. In one of the early 00:02:25.840 --> 00:02:33.519 training missions, they are running damn near identically, 00:02:30.879 --> 00:02:38.480 which is pretty unbelievable. And AMD hasn't even showed all their cards yet 00:02:35.760 --> 00:02:45.280 because their chip is running at 3.4 4 GHz without their precision boost 00:02:42.000 --> 00:02:48.519 dynamic clock speed adjustments going at 00:02:45.280 --> 00:02:48.519 all yet. 00:02:48.720 --> 00:02:56.319 Demo number two also highly controversial because people were 00:02:53.200 --> 00:03:00.879 looking at it going, "Well, how can this 00:02:56.319 --> 00:03:03.040 8 core versus that 8 core have so 00:03:00.879 --> 00:03:08.879 significant a difference in the stream quality?" And the answer from AMD and I 00:03:06.080 --> 00:03:12.720 I'm I mean I'm I'm standing here with me in the webcam after stream delay. I'm 00:03:11.040 --> 00:03:17.360 standing here seeing the demo. The answer from them is that it's their 00:03:14.800 --> 00:03:22.080 infinity fabric that allows the individual processing cores to talk to 00:03:19.920 --> 00:03:28.080 each other more efficiently that allows the game to be played here and the 00:03:24.640 --> 00:03:32.560 stream to be processed on the CPU for 00:03:28.080 --> 00:03:34.239 the broadcast much more efficiently. 00:03:32.560 --> 00:03:41.200 Now, the obvious question is why aren't you just using quicks sync or NBNC uh 00:03:38.080 --> 00:03:42.560 like a fixed function H.264 encoder and 00:03:41.200 --> 00:03:46.879 there are two answers to that. So, number one uh you quicksync doesn't 00:03:44.319 --> 00:03:52.319 exist on Intel's Broadwelli processors. They're they're high-end. Um, and number 00:03:49.280 --> 00:03:54.720 two is that image quality can be with 00:03:52.319 --> 00:03:58.959 the correct settings superior using CPU encoding. And they haven't toolled 00:03:56.159 --> 00:04:04.640 around with it in order to eliminate that advantage of CPU encoding. Again, 00:04:02.239 --> 00:04:09.840 very interesting. So, what does this mean for Ryzen? Uh there was a lot of 00:04:07.200 --> 00:04:13.519 speculation before KBL Lake dropped that AMD was putting it up against Broadwell 00:04:12.000 --> 00:04:17.759 E so that they had the advantage of comparing Ryzen to an Intel architecture 00:04:15.360 --> 00:04:21.199 that was effectively two generations old. But now that we've seen what 00:04:19.680 --> 00:04:25.440 Skylake brings to the table versus Broadwell, now that we've seen what KBL 00:04:22.960 --> 00:04:30.479 brings to the table IPC wise versus Skylake, this is shaping up to be a 00:04:28.400 --> 00:04:36.000 very, very interesting fight. and I can't wait for sometime Q120 2017 to get 00:04:33.520 --> 00:04:40.320 my hands on a real actual retail chip and a real retail board. I know I'm not 00:04:38.560 --> 00:04:43.680 the only one. So, thanks for checking out this video. Thanks to AMD for 00:04:42.000 --> 00:04:47.759 bringing us down here to the show and being a big part of our CES 2017 00:04:45.759 --> 00:04:53.720 coverage. And uh don't forget to subscribe so you don't miss any of our 00:04:49.120 --> 00:04:53.720 CES videos here at the show.