WEBVTT 00:00:00.240 --> 00:00:03.560 So back at Computex, when Intel did that weird demo 00:00:03.560 --> 00:00:05.760 on stage where they had this like, 00:00:05.760 --> 00:00:09.280 this like chilled liquid cooled 28 core 00:00:09.280 --> 00:00:11.660 overclocked processor, 00:00:11.660 --> 00:00:13.880 I don't think anyone could figure out 00:00:13.880 --> 00:00:17.760 what exactly it was they were trying to accomplish. 00:00:17.760 --> 00:00:21.080 Even Intel themselves didn't seem to know what was going on 00:00:21.080 --> 00:00:25.480 because that same CPU, although not cooled by chilled liquid 00:00:25.480 --> 00:00:28.980 was actually on display the next day 00:00:28.980 --> 00:00:31.820 with their board partners like ASUS, 00:00:31.820 --> 00:00:34.260 only to be quickly yanked away 00:00:34.260 --> 00:00:37.340 and then we haven't heard anything about it since then, 00:00:37.340 --> 00:00:41.740 except for a leaked roadmap that seems to suggest 00:00:41.740 --> 00:00:44.960 that Intel is planning what appears to be 00:00:44.960 --> 00:00:49.960 a knee jerk reaction to AMD's 32 core Threadripper 2, 00:00:50.980 --> 00:00:55.080 a 28 core based on as far as we can tell, 00:00:55.080 --> 00:00:58.840 a similar platform to their existing Xeon Platinum. 00:00:58.840 --> 00:01:01.180 Now, I don't know what segment they're gonna target 00:01:01.180 --> 00:01:04.060 with this, a workstation, enthusiast, 00:01:04.060 --> 00:01:06.100 people with more money than brains, 00:01:06.100 --> 00:01:09.220 but whatever it is, I realized something. 00:01:09.220 --> 00:01:12.040 We actually have in our hands, 00:01:12.040 --> 00:01:15.340 both the 32 core Threadripper 2 00:01:15.340 --> 00:01:19.540 and Intel Xeon Platinum 8180, 00:01:19.540 --> 00:01:23.000 which rumor has it will be very similar 00:01:23.000 --> 00:01:24.840 to the upcoming enthusiast chip, 00:01:24.840 --> 00:01:26.700 albeit without overclocking. 00:01:26.700 --> 00:01:28.840 So today we are gonna put them 00:01:28.840 --> 00:01:31.160 head to head as a preview 00:01:31.160 --> 00:01:35.020 of the ultimate heavyweight showdown of CPUs 00:01:35.020 --> 00:01:37.920 we're not sure who is going to buy. 00:01:37.920 --> 00:01:40.820 And the whole thing is brought to you by Videoblocks. 00:01:40.820 --> 00:01:43.380 Videoblocks is a great subscription-based resource 00:01:43.380 --> 00:01:46.260 for downloading stock footage, video, and more. 00:01:46.260 --> 00:01:57.790 Check it out at the link in the video description. 00:01:57.790 --> 00:01:59.730 So while Ivan puts the finishing touch 00:01:59.730 --> 00:02:02.870 on our LGA 3647 bench over there, 00:02:02.870 --> 00:02:05.390 I'm gonna walk you through what are otherwise 00:02:05.390 --> 00:02:06.890 as identical as we can get them to be. 00:02:06.890 --> 00:02:08.590 We can get them test rigs. 00:02:08.590 --> 00:02:12.130 So in here, we've got a Threadripper 32 core. 00:02:12.130 --> 00:02:15.250 We've got a Noctua 120 millimeter tower cooler. 00:02:15.250 --> 00:02:17.050 We've got a Zenith Xtreme motherboard 00:02:17.050 --> 00:02:21.050 with what I believe is 32 gigs of RAM, 32 gigs, right? 00:02:21.050 --> 00:02:21.890 Yeah, it should be. 00:02:21.890 --> 00:02:26.550 And then we've got a GTX 1080, Samsung 840 Pro SSD, 00:02:26.550 --> 00:02:29.510 and an RM850i power supply. 00:02:29.510 --> 00:02:31.570 Now, bear in mind that any results 00:02:31.570 --> 00:02:33.870 that we show you guys for this guy 00:02:33.870 --> 00:02:36.770 have to be taken with a significant grain 00:02:36.770 --> 00:02:39.710 of salt because Intel will have at least, I think, 00:02:39.710 --> 00:02:43.010 a couple of months to tune the consumer version of this 00:02:43.010 --> 00:02:45.310 before you would ever be able to buy one. 00:02:45.310 --> 00:02:47.670 Not that most people would ever be able to buy one. 00:02:47.670 --> 00:02:49.270 Like, this is 10 grand. 00:02:49.270 --> 00:02:51.610 So we don't know how much or how little they're planning 00:02:51.610 --> 00:02:53.370 to slash the price compared to the Pro one, 00:02:53.370 --> 00:02:55.790 but like, it ain't gonna be cheap. 00:02:55.790 --> 00:03:00.090 So this guy right here has got same power supply, 00:03:00.090 --> 00:03:03.530 same cooler, sort of, except it's like a modified version 00:03:03.530 --> 00:03:05.890 that works with LGA 3647. 00:03:05.890 --> 00:03:06.570 Thank you, Noctua, for that. 00:03:06.570 --> 00:03:07.750 Thank you, Noctua, for sending that over. 00:03:07.750 --> 00:03:10.150 We've got them both hooked up to power meters, 00:03:10.150 --> 00:03:14.690 and then it's running an ASUS Z11PA-U12. 00:03:14.690 --> 00:03:18.210 It's got the same GeForce GTX 1080 Founder's Edition, 00:03:18.210 --> 00:03:21.050 and then it is, of course, running a Xeon Platinum 8180. 00:03:21.050 --> 00:03:24.310 Now, the RAM, this is something that could affect our results, 00:03:24.310 --> 00:03:27.090 but we had to use what's compatible with our CPU. 00:03:27.090 --> 00:03:31.210 So this is DDR4-2666 ECC memory, 00:03:31.210 --> 00:03:36.730 the same stuff from our six workstations, one CPU project. 00:03:36.730 --> 00:03:38.300 Hey, it's done. 00:03:38.300 --> 00:03:41.730 Here we go. 00:03:41.730 --> 00:03:43.320 Is this broken? 00:03:43.320 --> 00:03:45.780 This CPU, what, what, what, what, what, what, what? 00:03:45.780 --> 00:03:48.820 Okay, adding another boulder-sized grain of salt 00:03:48.820 --> 00:03:53.020 to this whole experience is that this latest BIOS 00:03:53.020 --> 00:03:57.780 we just flashed no longer supports the 8180 CPU, 00:03:57.780 --> 00:03:59.260 probably, if I had to guess, 00:03:59.260 --> 00:04:03.960 because of its massive 205-watt TDP. 00:04:03.960 --> 00:04:08.000 So we're gonna have to go back in time to another BIOS 00:04:08.000 --> 00:04:10.000 in order to run our comparison. 00:04:11.960 --> 00:04:16.200 And four hours later, we finally have the solution. 00:04:16.200 --> 00:04:17.020 You know what? 00:04:17.020 --> 00:04:17.940 Screw it. 00:04:17.940 --> 00:04:19.180 We're just gonna run our... 00:04:22.050 --> 00:04:25.050 We're just gonna run the light of a thousand suns 00:04:25.050 --> 00:04:27.150 directly into Ivan's eyes. 00:04:27.150 --> 00:04:28.950 Spawning into my soul. 00:04:28.950 --> 00:04:29.790 Sorry about that. 00:04:29.790 --> 00:04:32.770 Anyway, none of the BIOSes we tried worked, 00:04:32.770 --> 00:04:34.970 and we downloaded, like, one of the earliest ones 00:04:34.970 --> 00:04:36.390 that should have been the same as what we had, 00:04:36.390 --> 00:04:38.510 and it was corrupt right off AC system. 00:04:38.510 --> 00:04:39.790 Whatever, it doesn't matter. 00:04:39.790 --> 00:04:41.870 Point is, we're just using our dual-socket board, 00:04:41.870 --> 00:04:43.670 with a single CPU, 00:04:43.670 --> 00:04:46.550 where the only real consideration performance-wise 00:04:46.550 --> 00:04:48.770 is we have to make sure we use a PCI Express slot 00:04:48.770 --> 00:04:50.530 that is connected to that socket. 00:04:50.530 --> 00:04:51.630 So everything's hunky-dory. 00:04:51.630 --> 00:04:54.230 This is effectively the same 00:04:54.230 --> 00:04:56.510 apples-ish to apples-ish comparison, 00:04:56.510 --> 00:04:58.490 and we are ready to go. 00:04:58.490 --> 00:05:01.350 So naturally, I'm sure the first thing you guys 00:05:01.350 --> 00:05:04.870 are gonna wanna see after the task manager porn, 00:05:04.870 --> 00:05:09.330 56 threads on this one, 64 threads on this one, 00:05:09.330 --> 00:05:10.670 is probably gonna be Cinebench, right? 00:05:10.670 --> 00:05:13.030 It's gotta be Cinebench. 00:05:13.030 --> 00:05:15.170 Boom! 00:05:15.170 --> 00:05:18.670 Who will finish first? 00:05:18.670 --> 00:05:20.610 Wow, those are both super fast, 00:05:20.610 --> 00:05:24.310 but one of them looks a little faster. 00:05:24.310 --> 00:05:26.370 One of them looks a fair bit faster. 00:05:26.370 --> 00:05:27.210 Whoa! 00:05:28.700 --> 00:05:32.720 AMD is a good 20% ahead, which makes sense, 00:05:32.720 --> 00:05:37.720 because even though Threadripper 2's performance per clock 00:05:37.860 --> 00:05:41.520 is lower than Intel's Skylake architecture, 00:05:41.520 --> 00:05:43.480 they do have a core advantage, 00:05:43.480 --> 00:05:46.860 so they have four more cores, and watch this. 00:05:46.860 --> 00:05:49.140 We're gonna go ahead, we're gonna run the test again, 00:05:49.140 --> 00:05:51.000 and we're gonna have a look at what our clock speeds are 00:05:51.000 --> 00:05:52.040 while they're going. 00:05:52.040 --> 00:05:55.940 So you can see here, AMD is turboing up higher 00:05:55.940 --> 00:05:59.590 when you're hitting all the cores compared to Intel. 00:05:59.590 --> 00:06:05.100 So round one goes to the Threadripper 2990WX. 00:06:05.100 --> 00:06:10.390 Test number two, the V-Ray benchmark from Chaos Group. 00:06:10.390 --> 00:06:14.400 So we're running the CPU benchmark here. 00:06:14.400 --> 00:06:16.760 Here we go, our AMD processor is pinned, 00:06:16.760 --> 00:06:19.120 our Intel processor is pinned. 00:06:19.120 --> 00:06:23.040 AMD's looking pretty good here too, wow. 00:06:23.040 --> 00:06:26.290 And done. 00:06:26.290 --> 00:06:29.410 Okay, no matter which one of these things you buy, 00:06:29.410 --> 00:06:33.140 you're gonna be like going super fast, 00:06:33.140 --> 00:06:37.040 and three seconds doesn't seem like a lot of difference, 00:06:37.040 --> 00:06:42.040 but three seconds out of 30 is 10% faster for Threadripper 2. 00:06:43.410 --> 00:06:44.810 This is great. 00:06:44.810 --> 00:06:46.150 Tell us more about your gear. 00:06:46.150 --> 00:06:47.070 Are you overclocking? 00:06:47.070 --> 00:06:48.730 Is it liquid cooling or anything else ingenious 00:06:48.730 --> 00:06:51.260 for getting a performance boost? 00:06:51.260 --> 00:06:53.540 Does it show that every time? 00:06:53.540 --> 00:06:56.340 Or is it just when you're like bananas? 00:06:56.340 --> 00:07:01.410 Hey Anthony, you ever seen this in the V-Ray benchmark? 00:07:01.610 --> 00:07:02.890 It did it on both of these. 00:07:02.890 --> 00:07:04.910 I wonder if there's some kind of a threshold. 00:07:04.910 --> 00:07:07.330 Clearly that's a tailored message. 00:07:07.330 --> 00:07:09.160 Yeah. 00:07:09.160 --> 00:07:10.620 No, we're not doing anything ingenious, 00:07:10.620 --> 00:07:13.120 just spending lots of money. 00:07:13.120 --> 00:07:14.720 So for our blender test, 00:07:14.720 --> 00:07:17.760 we're going to be rendering the classroom scene. 00:07:18.120 --> 00:07:21.180 And this one will probably actually take a couple minutes. 00:07:21.180 --> 00:07:22.640 So we'll come back in just a moment here. 00:07:22.640 --> 00:07:24.000 Man, that's a lot of cores. 00:07:24.000 --> 00:07:27.240 It's beautiful. 00:07:27.240 --> 00:07:30.300 You can even just feel the heat coming off these things. 00:07:30.300 --> 00:07:32.120 This does seem like a good opportunity 00:07:32.120 --> 00:07:33.600 to remind you guys though, 00:07:33.600 --> 00:07:36.440 that this is not the actual product. 00:07:36.440 --> 00:07:39.840 This is just the closest thing that we have to it now. 00:07:39.840 --> 00:07:40.860 This is cool though, 00:07:40.860 --> 00:07:43.400 compared to that burn test we were running before, 00:07:43.400 --> 00:07:46.380 it's turboing much higher in real world applications, 00:07:46.380 --> 00:07:48.420 even though the usage is a hundred percent. 00:07:48.420 --> 00:07:51.000 So it looks like they might be detecting 00:07:51.000 --> 00:07:53.380 stress testing applications and throttling down. 00:07:53.380 --> 00:07:56.500 Anyway, so that's interesting to know. 00:07:56.500 --> 00:07:57.860 In terms of power consumption, 00:07:57.860 --> 00:08:00.800 this is one area where Intel's looking a little bit better. 00:08:00.800 --> 00:08:04.100 Our Threadripper 2 machine is pulling 380 watts 00:08:04.100 --> 00:08:04.940 from the wall. 00:08:04.940 --> 00:08:07.320 Remember, this is before we've even touched the GPU, 00:08:07.320 --> 00:08:09.620 and this is with no overclocking on either of them, 00:08:09.620 --> 00:08:11.020 because our Xeon won't overclock, 00:08:11.020 --> 00:08:13.280 whereas we don't know what Intel's planning to do 00:08:13.280 --> 00:08:15.020 with the real chip that's coming. 00:08:15.020 --> 00:08:18.380 And then Intel's pulling about 265 watts. 00:08:18.620 --> 00:08:19.460 It's crazy. 00:08:19.460 --> 00:08:22.080 They both feel like space heaters, you know? 00:08:22.080 --> 00:08:25.590 With the fans, even with these fans running at full speed, 00:08:25.590 --> 00:08:27.050 and these are nice fans. 00:08:27.050 --> 00:08:32.440 Oh boy, this is not looking amazing for Intel right now. 00:08:33.900 --> 00:08:37.340 Threadripper 2 already chewed through 00:08:37.340 --> 00:08:42.420 the first segment of the test, and I think it's done. 00:08:42.420 --> 00:08:45.970 Wow. 00:08:45.970 --> 00:08:49.840 Just under five minutes for Classroom, 00:08:49.840 --> 00:08:51.700 whereas Intel, 00:08:51.700 --> 00:08:54.210 we're at 515, 00:08:54.210 --> 00:08:57.810 and it hasn't even finished the initial render. 00:08:57.810 --> 00:09:03.090 545. 00:09:03.090 --> 00:09:07.650 I don't think I ever thought I would see it again. 00:09:07.650 --> 00:09:12.330 AMD's top of the line versus Intel's top of the line, 00:09:12.330 --> 00:09:17.480 and it just being a blowout. 00:09:17.480 --> 00:09:20.120 So that concludes the productivity segment of our testing, 00:09:20.120 --> 00:09:25.080 and now we're gonna do what neither Intel nor AMD 00:09:25.080 --> 00:09:27.420 is gonna be particularly stoked on. 00:09:27.680 --> 00:09:29.080 We're gonna fire up some games. 00:09:29.080 --> 00:09:31.900 So starting with CSGO. 00:09:31.900 --> 00:09:34.940 Well, if CSGO is anything to go by, 00:09:34.940 --> 00:09:37.560 Intel looks like they're gonna have a bit better of a time 00:09:37.560 --> 00:09:38.660 when it comes to gaming. 00:09:38.660 --> 00:09:41.960 With that said, CSGO is known to heavily favor 00:09:41.960 --> 00:09:46.920 Intel's current chips versus AMD's Zen-based chips. 00:09:46.920 --> 00:09:47.900 Oh boy. 00:09:47.900 --> 00:09:53.750 Average frame rate, 310.21 00:09:53.750 --> 00:09:56.330 versus 297.85. 00:09:56.330 --> 00:09:58.460 That doesn't seem right. 00:09:58.460 --> 00:09:59.300 Because in game- 00:09:59.300 --> 00:10:02.060 This was so much higher. 00:10:02.060 --> 00:10:03.800 Hex? 00:10:03.800 --> 00:10:04.760 I have an alibi. 00:10:04.760 --> 00:10:06.720 Yeah, I don't know what just happened there. 00:10:06.720 --> 00:10:08.280 There's like, there's no way. 00:10:08.280 --> 00:10:11.620 Okay, well take those results with even more salt 00:10:11.620 --> 00:10:13.260 than the rest of this video, 00:10:13.260 --> 00:10:14.920 because there's no way that's right. 00:10:14.920 --> 00:10:16.970 That doesn't make any sense. 00:10:16.970 --> 00:10:18.330 All right. 00:10:18.330 --> 00:10:21.500 Our test is done. 00:10:21.500 --> 00:10:24.180 We can go ahead and quit the game. 00:10:24.180 --> 00:10:25.020 Okay. 00:10:25.020 --> 00:10:26.260 So these are more like the numbers 00:10:26.260 --> 00:10:28.780 we were expecting to see in CSGO. 00:10:28.780 --> 00:10:29.880 That brings us to our- 00:10:29.880 --> 00:10:31.260 Our final game. 00:10:31.260 --> 00:10:32.820 Shadow of the Tomb Raider. 00:10:32.820 --> 00:10:34.700 How to benchmark on the same account 00:10:34.700 --> 00:10:37.530 with two PCs at once. 00:10:37.530 --> 00:10:40.530 You gotta do the Ethernet cable shuffle. 00:10:40.530 --> 00:10:41.510 It goes like this. 00:10:45.260 --> 00:10:48.780 Play game. 00:10:48.780 --> 00:10:52.360 Here we go. 00:10:52.360 --> 00:10:54.340 Now this is closer. 00:10:54.340 --> 00:10:56.020 This is really close. 00:10:56.020 --> 00:10:58.120 With all these character models, 00:10:58.120 --> 00:11:02.620 Intel is actually enjoying a pretty significant advantage. 00:11:02.620 --> 00:11:04.200 Average FPS, 100. 00:11:04.200 --> 00:11:06.000 Average FPS, 87. 00:11:06.000 --> 00:11:07.080 So, a significant advantage. 00:11:07.080 --> 00:11:12.500 But it should be noted that a gaming advantage 00:11:14.620 --> 00:11:16.280 from one of these CPUs to the other 00:11:16.280 --> 00:11:18.200 is not particularly meaningful. 00:11:18.200 --> 00:11:19.880 If you wanted to game, 00:11:19.880 --> 00:11:24.320 you would run out and you would buy a secondhand 7700K 00:11:24.320 --> 00:11:26.220 and overclock the snot out of it. 00:11:26.220 --> 00:11:28.380 Oh right, thermals. 00:11:28.380 --> 00:11:29.840 We're not done yet. 00:11:29.840 --> 00:11:31.320 So we already have a pretty good idea 00:11:31.320 --> 00:11:32.400 of what we're gonna see here, 00:11:32.400 --> 00:11:35.200 simply because the power consumption 00:11:35.200 --> 00:11:37.500 on Intel's platform was so much lower. 00:11:37.720 --> 00:11:40.040 Yeah, so this is consistent with what we saw before 00:11:40.040 --> 00:11:43.680 with Intel's cores coming in about five degrees lower. 00:11:43.680 --> 00:11:46.340 So there you have it, guys. 00:11:46.340 --> 00:11:48.220 I said it before, but it bears repeating. 00:11:48.220 --> 00:11:50.820 Take everything you saw here with a grain of salt 00:11:50.820 --> 00:11:52.860 because we don't know what, if anything, 00:11:52.860 --> 00:11:55.300 Intel will change about the real product 00:11:55.300 --> 00:11:57.280 that they're supposedly coming out with. 00:11:57.280 --> 00:11:59.160 I mean, something like an increase 00:11:59.160 --> 00:12:01.580 to the all-core boost frequencies 00:12:01.580 --> 00:12:03.980 could make a significant difference 00:12:03.980 --> 00:12:05.980 to the results that we got today. 00:12:05.980 --> 00:12:07.480 But what it looks like? 00:12:07.480 --> 00:12:08.360 But what it looks like? 00:12:08.360 --> 00:12:13.360 For now, is that for the first time in over a decade, 00:12:14.180 --> 00:12:18.900 in a pure who's got the biggest wheelie competition, 00:12:18.900 --> 00:12:20.880 because I don't know that we would necessarily 00:12:20.880 --> 00:12:24.660 put a two-thumbs-up recommendation behind either of these, 00:12:24.660 --> 00:12:26.990 depending on pricing, 00:12:26.990 --> 00:12:32.680 AMD might have the bigger wheelie. 00:12:32.680 --> 00:12:35.340 Who'da thunk it? 00:12:35.340 --> 00:12:36.960 And this wheelie measuring contest 00:12:36.960 --> 00:12:39.960 was brought to you, of course, by Videoblocks. 00:12:39.960 --> 00:12:41.080 Videoblocks is the place 00:12:41.080 --> 00:12:44.440 to get studio-quality footage for a fraction of the cost, 00:12:44.440 --> 00:12:45.900 and we use it all the time, 00:12:45.900 --> 00:12:48.020 particularly on our Techquickie channel. 00:12:48.020 --> 00:12:49.400 With Videoblocks, you can download 00:12:49.400 --> 00:12:51.160 all the stock video your heart desires 00:12:51.160 --> 00:12:52.260 from their member library, 00:12:52.260 --> 00:12:54.740 including HD and 4K footage, 00:12:54.740 --> 00:12:57.500 After Effects templates, motion backgrounds, and more. 00:12:57.500 --> 00:12:59.780 Plus, you can get exclusive discounts 00:12:59.780 --> 00:13:02.400 on millions of additional Marketplace clips 00:13:02.400 --> 00:13:05.700 where you save 40% on every purchase compared to non-members, 00:13:05.700 --> 00:13:08.400 and the original artists take home a commission 00:13:08.400 --> 00:13:10.100 of the sale price. 00:13:10.100 --> 00:13:10.900 All the content is royalty free, so if you're interested, 00:13:10.900 --> 00:13:14.020 you can use it for commercial and personal projects, 00:13:14.020 --> 00:13:15.260 such as YouTube videos, 00:13:15.260 --> 00:13:17.120 and new clips get added regularly, 00:13:17.120 --> 00:13:19.140 so there's always something new to check out. 00:13:19.140 --> 00:13:22.760 So head to the link in the video description to learn more. 00:13:22.760 --> 00:13:23.700 So thanks for watching, guys. 00:13:23.700 --> 00:13:24.960 If you disliked this video, 00:13:24.960 --> 00:13:27.540 then clearly you're not into, like, 00:13:27.540 --> 00:13:29.560 you know, heavyweight title fights, 00:13:29.560 --> 00:13:31.660 because that's basically what we saw today. 00:13:31.660 --> 00:13:33.520 But if you did like it, hit that like button, 00:13:33.520 --> 00:13:35.200 get subscribed, maybe consider checking out 00:13:35.200 --> 00:13:37.520 where to buy the stuff we featured 00:13:37.520 --> 00:13:39.380 at the link in the video description. 00:13:39.380 --> 00:13:40.740 Just like, you know, 00:13:40.740 --> 00:13:42.620 you open up your wallet, 00:13:42.620 --> 00:13:45.020 you start, you know, pulling out the cash, 00:13:45.020 --> 00:13:47.000 and then the credit cards, you know what? 00:13:47.000 --> 00:13:49.860 Forget it, just take the whole thing, right? 00:13:49.860 --> 00:13:51.860 So what else was I gonna say? 00:13:51.860 --> 00:13:52.940 Right, our merch store. 00:13:52.940 --> 00:13:53.980 You can check that out at the link 00:13:53.980 --> 00:13:54.820 in the video description. 00:13:54.820 --> 00:13:56.560 Also, we have our community forum down there, 00:13:56.560 --> 00:13:57.960 which you should totally join.