WEBVTT 00:00:00.000 --> 00:00:05.640 This is how I used to eat breakfast. And this is how I eat breakfast now. 00:00:05.640 --> 00:00:09.560 See the difference? Thanks to the Ryzen 7 9850X3D, 00:00:09.560 --> 00:00:12.800 I now do everything 7.69% faster, 00:00:12.800 --> 00:00:17.320 which doesn't sound like a big deal, but as far as mid-generation refreshes go, 00:00:17.320 --> 00:00:22.400 this one is pretty all right. TLDR, it costs a few dollars more for a chip 00:00:22.400 --> 00:00:25.680 that is by and large the same as the 9800X3D 00:00:25.680 --> 00:00:29.320 that you already know and love, except it boosts its clock speeds higher, 00:00:29.320 --> 00:00:33.240 resulting in measurable gains across productivity and eSports gaming. 00:00:33.240 --> 00:00:36.800 And it does so without running hotter or louder 00:00:36.800 --> 00:00:40.720 or drawing a ton more power. Is it the most exciting chip launch 00:00:40.720 --> 00:00:43.960 we've ever seen from AMD? No, but it never claimed to be. 00:00:43.960 --> 00:00:48.200 It's just some mid-generation fine wine aging for the AM5 platform, 00:00:48.200 --> 00:00:52.680 while we wait for a true successor based on Zen 6 toward the end of this year or early next. 00:00:52.680 --> 00:00:58.760 So with that in mind, let's take a look at performance to see who should be interested in the 9850X3D 00:00:58.760 --> 00:01:03.240 and maybe more importantly, who should be holding on to their existing hardware 00:01:03.240 --> 00:01:06.640 and saving their money. And also, who should be checking out our sponsor? 00:01:06.640 --> 00:01:09.480 ["RTX 5090s"] 00:01:15.480 --> 00:01:18.680 RTX 5090s are in short supply, even here at LTT. 00:01:18.680 --> 00:01:22.400 So we set up four identical benches with RTX 4090s 00:01:22.400 --> 00:01:27.520 and got right to work on gaming performance where AMD's claims were immediately validated. 00:01:27.520 --> 00:01:32.800 The king got a little king glir with the 9850X3D leading the pack 00:01:32.800 --> 00:01:36.880 across our entire suite of eSports titles and 1080p benchmarks. 00:01:36.880 --> 00:01:41.760 Now I could spend the rest of my slideshow presentation here talking about the cases where it leads by a little bit 00:01:41.760 --> 00:01:47.000 and the ones where it leads by a little bit more. But instead, I'm simply gonna say it wins 00:01:47.000 --> 00:01:52.000 and I'm gonna use the rest of my time to explain why reviewers even bother to test like this. 00:01:52.000 --> 00:01:55.640 I mean, surely no one buying a $500 gaming CPU 00:01:55.640 --> 00:01:59.520 is gonna be running at 1080p low on an RTX 4090, right? 00:01:59.520 --> 00:02:05.480 Well, yeah, you're right. But we do this because we wanna know which chip is faster, period. 00:02:05.480 --> 00:02:09.080 Not which one handles better, not which one is more likely to impress the chicks 00:02:09.080 --> 00:02:14.560 at the club, which one is faster. And when you wanna know which chip pumps out frames better, 00:02:14.560 --> 00:02:17.840 lowering both your resolution and your in-game details 00:02:17.840 --> 00:02:21.360 helps to isolate the CPU's game engine handling performance 00:02:21.360 --> 00:02:26.400 and really separate the fastest chip from the almost fastest chip. 00:02:26.400 --> 00:02:30.060 And AMD's 9850X3D is the fastest gaming chip 00:02:30.060 --> 00:02:35.080 that we tested, thanks to all of the things that made its predecessor processor the fastest before. 00:02:35.080 --> 00:02:38.680 Eight of AMD's Zen 5 cores coupled with ample 3DV cache 00:02:38.680 --> 00:02:44.760 creates a combo that puts both Intel's latest Core Ultra desktop chips and their older 14th gen flagship 00:02:44.760 --> 00:02:48.800 to absolute shame. But does any of this matter? 00:02:48.800 --> 00:02:53.760 To answer that, we need to get a little more realistic, both in terms of in-game visuals 00:02:53.760 --> 00:02:56.800 and the real world use cases for these chips. 00:02:56.800 --> 00:03:02.160 Let's step up then to 1440P resolution and throw in some tests with ray tracing where, 00:03:02.160 --> 00:03:06.640 whoa, the 9850X3D no longer sits alone atop the roost, 00:03:06.640 --> 00:03:11.000 king of his domain, but becomes simply another runner in the race. 00:03:11.000 --> 00:03:14.280 Still a very fast chip boasting the highest average FPS 00:03:14.280 --> 00:03:19.160 in both Counter-Strike 2 and Rocket League. So it's clear then that once we start 00:03:19.160 --> 00:03:24.000 to introduce more variables, the story becomes muddier to say the least 00:03:24.000 --> 00:03:29.800 with certain games favoring some CPU architectures more than others and those advantages shrinking 00:03:29.800 --> 00:03:35.200 the more GPU eye candy you turn on with ray tracing being the ultimate equalizer, 00:03:35.200 --> 00:03:39.760 bringing every chip we tested within about 5% of the rest. 00:03:39.760 --> 00:03:43.120 Which isn't to say though, that you should just buy any chip 00:03:43.120 --> 00:03:47.720 and expect exactly the same experience. In the longer term, theoretically, 00:03:47.720 --> 00:03:51.680 a faster gaming chip will last you through more GPU upgrades, 00:03:51.680 --> 00:03:57.400 extending the time before your next platform upgrade. And with upscaling technology becoming more 00:03:57.400 --> 00:04:02.520 and more important all the time, rendering high frame rates at lower resolutions 00:04:02.520 --> 00:04:08.200 and then upscaling them with AI could become increasingly important for high-end gaming. 00:04:08.200 --> 00:04:11.760 Okay then Linus, so just buy the one at the top of the chart then. 00:04:11.760 --> 00:04:17.240 Well, not that either. It's horses for courses. If all that matters to you is the highest FPS 00:04:17.240 --> 00:04:23.040 at lower medium details or any sports titles. Yeah, it doesn't get any better than the 9850X3D, 00:04:23.040 --> 00:04:28.200 but for almost anyone else, we need to at least think about the bigger picture. 00:04:28.200 --> 00:04:31.480 What about whole platform pricing? Where could I score a deal? 00:04:31.480 --> 00:04:36.000 Intel's older generation 12th gen CPUs are still potent gaming chips 00:04:36.000 --> 00:04:40.080 and they support cheaper DDR4 memory. We'll have a full video on that coming soon, 00:04:40.080 --> 00:04:43.120 but the TLDR is, it's a pretty compelling value. 00:04:43.120 --> 00:04:49.960 Or if I'm willing to go to the second hand market, 5700X3D pricing has not spiked nearly as high 00:04:49.960 --> 00:04:54.400 as 5800X3Ds and AM4 motherboards and matching DDR4 00:04:54.400 --> 00:04:58.440 can be had for cheap at an open box 1440P gaming monitor. 00:04:58.440 --> 00:05:04.920 And that's an entire platform refresh for the price of a 9850X3D CPU alone, 00:05:04.920 --> 00:05:09.600 which would leave plenty of budget to get some nice new cables at lttstore.com. 00:05:09.600 --> 00:05:14.760 We'll have that link down below. Of course, none of what I just said means the 9850X3D is bad 00:05:14.800 --> 00:05:19.840 or a bad deal, it's bad CPU. No, it's not. The best of the best always comes at a premium 00:05:19.840 --> 00:05:23.240 and honestly, it's a better product than I thought it was gonna be. 00:05:23.240 --> 00:05:28.680 My initial reaction was, ah, it's probably gonna be a righted out 9800X3D. 00:05:28.680 --> 00:05:32.100 And I expected a big boost to both power draw 00:05:32.100 --> 00:05:38.200 and heat output in order for AMD to reach their lofty 5.6 gigahertz boost frequency targets. 00:05:38.200 --> 00:05:41.800 But that turned out to not be the case. In our Cinebench power draw test, 00:05:41.800 --> 00:05:45.440 AMD's finest hit higher clock speeds than its predecessor, 00:05:45.440 --> 00:05:50.960 while running ever so slightly cooler and drawing ever so slightly less power. 00:05:50.960 --> 00:05:54.040 We didn't see quite the same thing in F124. 00:05:54.040 --> 00:05:59.280 Power draw was actually up a little bit on average. But still, both of these results are so close 00:05:59.280 --> 00:06:03.560 that it is safe to say that this is not a meaningfully more power hungry chip 00:06:03.560 --> 00:06:08.000 and it does clock noticeably higher. This translates to a better experience 00:06:08.000 --> 00:06:13.080 outside of games as well, where the 9850X3D consistently beats its 50-less cousin 00:06:13.120 --> 00:06:18.320 across both our productivity and our creative benchmarks with small victories in multi-threaded workloads 00:06:18.320 --> 00:06:23.280 and larger ones in single-threaded applications where it really gets to flex its boosted clock speed. 00:06:23.280 --> 00:06:28.120 It's worth noting, by the way, we didn't run our entire suite here since these X3D chips really are marketed 00:06:28.120 --> 00:06:31.160 primarily towards gaming, but we still wanted to show a snapshot 00:06:31.160 --> 00:06:35.400 of what they're capable of outside of games. Now, I'm not gonna go into this too much, 00:06:35.400 --> 00:06:39.700 but Intel's core count advantage translates into some really lopsided results 00:06:39.700 --> 00:06:42.920 in our multi-threaded benchmarks. And that's what their last-gen Core Ultra 00:06:42.920 --> 00:06:46.920 can't wait to see what happens when their new architecture makes its way from mobile to the desktop. 00:06:46.920 --> 00:06:51.160 Core Ultra next-gen versus Zen 6 is gonna be a heck of a battle. 00:06:51.160 --> 00:06:57.400 Of course, that's a story for tomorrow. The story for today is Ryzen 7 9850X3D. 00:06:57.400 --> 00:07:01.560 It's a little faster and it costs a little more. Not the most exciting chip, 00:07:01.560 --> 00:07:04.680 but certainly better than AMD simply resting on their laurels 00:07:04.680 --> 00:07:08.840 and waiting for Intel to close the gap. And I was really impressed to see that their manufacturing 00:07:08.840 --> 00:07:14.360 has improved so much in the last year that they can do this without juicing power consumption. 00:07:14.360 --> 00:07:18.240 Now, to juice this segue to our sponsor. If you guys enjoyed this video, 00:07:18.240 --> 00:07:24.000 why not check out our recent look at AMD's top of the top of the top of the line Workstation Pro chip. 00:07:24.000 --> 00:07:29.640 That thing is nuts. It costs more than an entire PC with this and a 5090 in it. 00:07:29.640 --> 00:07:30.680 Just a CPU.