WEBVTT 00:00:00.000 --> 00:00:08.440 AMD, I need you to try harder. I mean, it's not like you're actively antagonizing your partner's customers in the media. 00:00:08.440 --> 00:00:16.960 That's more NVIDIA's jam. It's just that if you want to have your Ryzen moment in GPUs, now is the time to find 00:00:16.960 --> 00:00:22.920 the hunger that turned you from A player in the CPU space to the A player that you've 00:00:22.920 --> 00:00:26.160 become. I mean, don't get me wrong. 00:00:26.160 --> 00:00:33.000 The new Radeon 9060XT looks pretty good compared to NVIDIA's lukewarm 5060 series, and that 00:00:33.000 --> 00:00:37.760 might win you some sales, but I want you to look at the big picture. 00:00:37.760 --> 00:00:43.440 This is an opportunity while your main competitor is distracted by their rapidly growing AI 00:00:43.440 --> 00:00:50.760 business to change the perception of Radeon from that brand for scrub gamers to that brand 00:00:50.760 --> 00:00:54.080 that I never knew how much I liked until I tried it. 00:00:54.080 --> 00:00:59.400 It's an opportunity to build an army of loyal Radeon customers who, through their sheer 00:00:59.400 --> 00:01:04.600 numbers, will incentivize your board partners to build exciting solutions and incentivize 00:01:04.600 --> 00:01:09.720 developers to focus on Radeon when they're optimizing their games and software. 00:01:09.720 --> 00:01:16.360 But I'm getting ahead of myself a little bit. Adam just walked out of the keynote and he's got the details on everything from the new 00:01:16.360 --> 00:01:20.800 Radeon cards to AMD's fastest thread rippers ever. 00:01:20.800 --> 00:01:24.560 He's also got a segue to our sponsor. 00:01:24.560 --> 00:01:30.400 The RX 9060XT is AMD's attempt to appeal to gamers who reject NVIDIA's lackluster 8 00:01:30.400 --> 00:01:37.640 Gigabyte cards while also offering a profit margin-friendly 8 Gigabyte option for shareholders. 00:01:37.640 --> 00:01:41.920 To their credit, unlike Team Green, who has a history of quietly downgrading their lower 00:01:41.920 --> 00:01:47.720 VRAM variants in other ways, AMD has built both of these new cards using the same Navi 00:01:47.720 --> 00:01:55.240 44 GPU, with 32 compute units, up to 3.1 GHz boost clocks, and up to 821 AI tops running 00:01:55.240 --> 00:01:59.240 through a 16-lane PCIe Gen 5 interface. 00:01:59.240 --> 00:02:04.720 Thank you, AMD, for understanding that budget cards often end up in budget systems with 00:02:04.720 --> 00:02:11.080 older Gen PCIe slots, and having those extra lanes makes a big difference in compensating 00:02:11.080 --> 00:02:18.800 for those lower speeds. Power targets are 150 and 180 watts for the 8 Gigabyte and 16 Gigabyte cards, respectively, 00:02:18.800 --> 00:02:27.680 and according to AMD, the 9060XT 16 Gigabyte will be the world's best GPU under $350. 00:02:27.680 --> 00:02:30.960 And I'd like to believe them because s*** is pretty dire for budget gamers right now, 00:02:30.960 --> 00:02:37.960 but there's a couple of red flags that I want to point out. AMD hasn't presented a clear path to ensure that the 9060XTs will be actually available 00:02:37.960 --> 00:02:43.640 at their MSRPs. And for another, I couldn't help but noting the way that AMD focused on comparisons between 00:02:43.640 --> 00:02:47.720 their 16 Gigabyte card and NVIDIA's 5060 Ti 8 Gigabyte. 00:02:47.720 --> 00:02:51.920 It's a valid comparison, and the numbers do look impressive, with AMD claiming a roughly 00:02:51.920 --> 00:02:58.000 15% better performance per dollar in raster and in ray tracing at 1440p, but how much 00:02:58.000 --> 00:03:04.400 of that margin comes from VRAM issues that will pop up on the 9060XT 8 Gigabyte as well? 00:03:04.400 --> 00:03:08.760 Also, what is that? Enhance. 00:03:08.760 --> 00:03:18.120 Is that just three display outputs? AMD at least did confirm during the press briefing that these are DisplayPort 2.1a UHBR13.5 00:03:18.120 --> 00:03:25.720 and HDMI 2.1b. So at least you'll be able to natively drive all but the highest bandwidth displays. 00:03:25.720 --> 00:03:31.120 And I guess if you really need more monitors, there's always DisplayPort multi-stream transport, 00:03:31.120 --> 00:03:35.640 but it just feels like a bit of a weird spot for them to cut costs. 00:03:35.640 --> 00:03:41.040 Although if it helped contribute to their competitive pricing, then maybe that's a good thing? 00:03:41.040 --> 00:03:50.240 Well, sort of. See the 9060XT 16 gig is looking pretty good, undercutting the MSRP of the 5060 Ti 16 gig 00:03:50.240 --> 00:03:54.840 and even the 8 gig by $80 and $30 respectively. 00:03:54.840 --> 00:04:00.760 But the 8 gig version of the 9060XT is stunningly non-competitive. 00:04:00.760 --> 00:04:05.640 Something that AMD tacitly acknowledges when they gave it approximately two seconds of 00:04:05.640 --> 00:04:08.080 screen time during their announcement. 00:04:14.080 --> 00:04:20.880 Yikes. I mean sure, they price match NVIDIA's 5060, but this is exactly what I was getting at 00:04:20.880 --> 00:04:27.720 during the intro. Price matching the guy that everyone hates for overcharging is not enough. 00:04:27.720 --> 00:04:32.800 And it doesn't seem like they're even trying to compete with Intel's B series GPUs. 00:04:32.800 --> 00:04:38.360 The whole thing even just feels like backwards progress compared to AMD's own lineup, with 00:04:38.360 --> 00:04:45.600 gamers expected to pay $30 more than last gen's disappointing 8 gig card, the RX 7600. 00:04:45.600 --> 00:04:50.360 Anyway, these cards will be launching on June 5th, so keep your eyes peeled for our upcoming 00:04:50.360 --> 00:04:56.120 coverage. While we don't know for sure if their prices will be fake, what we do know for sure is 00:04:56.240 --> 00:05:01.840 there will be fake frames. The latest version of FSR, AMD's AI upskating tech, brought massive improvements earlier 00:05:01.840 --> 00:05:07.320 this year and AMD plans to keep mining for extra performance with FSR Redstone, which 00:05:07.320 --> 00:05:14.080 focuses on improving path tracing workloads specifically. To demonstrate this, they showed off this unflattering demo that looks like it was ripped 00:05:14.080 --> 00:05:19.480 out of a Project Gotham Racing from Xbox 360 era, like, what is this? 00:05:19.480 --> 00:05:28.280 Why does it look like this? I thought it was a tech demo. Anyways, the updates are, one, neural radiance caching, which learns how light bounces within 00:05:28.280 --> 00:05:31.680 a scene to try and predict ray intersections before they happen. 00:05:31.680 --> 00:05:36.080 Two, ray regeneration, which is NVIDIA's ray reconstruction, but more AMD ear. 00:05:36.080 --> 00:05:41.200 Three, super resolution, which is super resolution. And four, frame gen, which is frame gen. 00:05:41.200 --> 00:05:46.400 Sounds cool. What would be cooler is having more games supporting the tech. 00:05:46.400 --> 00:05:50.200 Oh, uh, okay, what next? We've got AI, AI. 00:05:50.320 --> 00:05:54.120 Oh, right, a new Threadripper. The king is back, baby. 00:05:54.120 --> 00:05:57.680 And the new Threadripper 9000 lineup looks wild. 00:05:57.680 --> 00:06:02.920 While AMD isn't packing any more cores than last year, Shimada peak seems to squeeze a 00:06:02.920 --> 00:06:08.120 lot of juice out of the move to AMD's more efficient Zen 5 architecture, starting with 00:06:08.120 --> 00:06:14.600 a 300 megahertz higher boost clock, which doesn't sound like much, but when it's across 00:06:14.600 --> 00:06:18.080 this many cores, it adds up to a lot. 00:06:18.280 --> 00:06:23.040 Now, if you were hoping for 3DV cash on these models, though, unless we real, we were all 00:06:23.040 --> 00:06:26.440 hoping for 3DV cash, you're going to have to keep waiting. 00:06:26.440 --> 00:06:32.600 But despite the minor changes to the on paper specs, AMD showed off a pretty decent performance 00:06:32.600 --> 00:06:38.320 uplift over their last gen and Cinebench before they pivoted to shaming Intel for a while. 00:06:38.320 --> 00:06:42.320 You know what? Fair enough, AMD. Enjoy your victory lap. 00:06:42.320 --> 00:06:47.400 Now naturally, given that these are workstation chips, AMD is also emphasizing how improved 00:06:47.400 --> 00:06:52.000 CPU performance will allow for improved AI performance when you're working with large 00:06:52.000 --> 00:06:57.320 models that require multiple GPUs, which is becoming more and more relevant as larger 00:06:57.320 --> 00:07:04.800 local models are gaining popularity. The best part, though, is these new chips are drop in compatible with existing 7000 00:07:04.800 --> 00:07:08.360 series motherboards with just a BIOS update. 00:07:08.360 --> 00:07:12.960 Good guy AMD, finally delivering on a Threadripper upgrade path. 00:07:12.960 --> 00:07:17.560 Just like we delivered these great cargo pants, available in two inseam lengths for shorter 00:07:17.560 --> 00:07:20.680 and for taller kings. lttstore.com. 00:07:20.680 --> 00:07:26.600 There is bad news, though. If you want to get your fingers on a 96-core Threadripper pro when they arrive in July, 00:07:26.600 --> 00:07:31.100 they will only be available through partners like Lenovo, meaning that enthusiasts are 00:07:31.100 --> 00:07:36.040 going to be stuck with the non-pro lineup that caps out at just 64 cores with fewer 00:07:36.040 --> 00:07:41.560 PCIe lanes, half the number of memory channels, and far less supported memory capacity. 00:07:42.000 --> 00:07:46.640 AMD has well and truly forgotten the Threadripper enthusiasts who supported them in the 00:07:46.640 --> 00:07:49.640 early days of Ryzen. Cool. 00:07:49.640 --> 00:07:58.520 Also, there's no pricing information today. It really is amazing how much AMD acts like Intel and NVIDIA when they are the top dog. 00:07:58.520 --> 00:08:02.000 Also coming to the workstation is some new AMD GPU tech. 00:08:02.000 --> 00:08:06.160 Earlier this week we did a video on Intel's B60 GPU and we talked about why it was their 00:08:06.160 --> 00:08:13.520 most important GPU yet. Making an affordable option for entry-level workstations that required certified drivers 00:08:13.520 --> 00:08:18.400 and creating a low-cost platform for running pretty large AI models locally. 00:08:18.400 --> 00:08:22.960 Meanwhile, AMD is hoping to wriggle their way into the pro GPU market by throwing a 00:08:22.960 --> 00:08:30.560 whopping 32GB of VRAM on their new Radeon AI Pro R9700, a GPU name optimized for shareholder 00:08:30.560 --> 00:08:35.760 value. AMD is also making big strides to improve their software, bringing improved ROK-M support 00:08:35.760 --> 00:08:40.520 to Windows and Linux alike, which enables developers to bring their AI and HPC applications 00:08:40.520 --> 00:08:43.520 to Radeon. Get it away from CUDA is what they say. 00:08:43.520 --> 00:08:49.320 They say, please, please no CUDA, sir. They didn't say how much this GPU would cost, but they did spend a fair amount of time 00:08:49.320 --> 00:08:54.200 comparing it to the 5080, so a thousand? 00:08:54.200 --> 00:08:57.400 Probably not. It's probably not going to be a thousand. 00:08:57.400 --> 00:09:03.440 AMD's announcements this year then were a mixed bag of exciting and also underwhelming. 00:09:03.440 --> 00:09:08.400 I mean, compared to the NVIDIA keynote, at least there was something worth talking about. 00:09:08.400 --> 00:09:13.760 And we took this incredible GPU and we shrunk it in here. 00:09:13.760 --> 00:09:21.520 Does that make any sense? And AMD is clearly busting their buns to catch up to NVIDIA and the workstation and data 00:09:21.520 --> 00:09:25.600 center spheres, but like the rest of the industry, they seem to have forgotten about 00:09:25.600 --> 00:09:31.040 their customers who don't have six, seven, or eight-figure annual budgets. 00:09:31.040 --> 00:09:36.000 I guess we're all just chopped liver. You know what isn't chopped liver, though? 00:09:36.000 --> 00:09:41.640 Our sponsor. If you guys enjoyed this video, maybe check out the one we did on Intel's most disruptive 00:09:41.640 --> 00:09:44.480 GPU yet, the ARC Pro B60.