AMD Ryzen 3000 – A Very REAL Problem For Intel
So it’s been a little over a week since AMD showed off their upcoming processors and graphics cards, and honestly there is a lot to talk about. In fact, I’ve actually covered all of the Navi GPU stuff in a separate video that you can check out right over here. However, this article is a little bit different from our normal Explained series. We wanted to put a little bit more time into reading all of the information that AMD has bestowed upon us regarding Zen 2 and the upcoming Ryzen 3000 series processors. The further we got into the information, the same question came up over and over again: How does this all compare to what Intel has to offer right now and are Intel desktop platforms really in a bunch trouble. Let’s just cut through the hype and get into this from a more balanced perspective while I explain some of the finer points of AMD’s 3rd generation Ryzen.
All right, so let’s start with the obvious thing about this whole Intel versus AMD CPU wars. Team Blue was caught with their pants down when AMD announced Zen. Now what could even be worrying for Team Blue is that Zen 2 seems to be performing even better than anyone else expected, including AMD. They initially thought they could achieve a 7% to 10% uplift in instructions per clock or IPC, but right now it’s hovering around 15%. But there is more too, usually companies have some issues moving to a new manufacturing process, like Intel sticking to their 14nm for about half a decade now since the 10nm just wasn’t ready. But with the new 7nm technology AMD has been able to drastically increase density while also cutting power, and of course increasing performance per watt. Also do keep in mind that switching processes usually leads to decreased core frequencies between one generation to the next, at least initially while the process is perfected. However, somehow AMD was able to avoid all of this, and so their new 7nm base CPUs will have equal or higher frequencies than the ones that they replace. Now if Intel was hoping for lower frequencies to offset AMD’s architectural improvements, well they were clearly wrong. I also want to focus a bit on that density point from before, because it’s going to be really important in the battle against Intel and it’s a key point to understanding how AMD can do what they’re doing.
The 7nm Advantage
You might’ve noticed that Lisa Su showed off a Ryzen CPU that looked something like this, with two smaller dies alongside one larger one. Those small cores are what AMD calls chiplets and they happen to be the cornerstone of Ryzen’s future. And remember I said that density matters well, each of those little 74 square millimeter chips, houses processing cores, cache, and more. Meanwhile, the larger die you see here is the input/output section that has things like the Infinity Fabric, the memory controllers and device communications.
Why Zen 2 Chiplet Design Is HUGE
Digging in a bit deeper, those core complex designs chiplets or CCDs are super small, but the 7nm manufacturing process allows them to house two quad-core/eight-thread Ryzen cores and 32 megabytes of L3 cache. Now when you add two of these CCDs together and you have yourself a CPU with 16 cores and 32 threads. While other designs can be created by simply disabling cores. For example, a 12 core part can be created by including two core complex designs and disabling one quad core section.
Meanwhile, an 8-core/16-thread CPU can be created by including just a single one of these chiplets alongside the cIOD. The really cool thing here is that in theory these eight cores can keep being added or removed to create all kinds of different products. Now for Intel this could cause a massive challenge, since AMD can now use as modular approach to pack as many 7nm cores onto the CPU as they can fit. This will allow them to launch relatively affordable and efficient CPUs with higher thread counts that could potentially overcome some of Intel’s most expensive HEDT processors.
A perfect example of that would be the upcoming Ryzen 9 3950X, a 16-core/32-thread monster processor that will sell for around $750 USD. I can’t imagine the look of horror on Intel faces when that price was announced, since their closest competitor would be the Core i9-9960X that goes for a ridiculous $1700 USD.
How Thermals May Be An Issue
Now the new chiplet approach does bring to mind something potentially concerning and that is heat. You see while the Ryzen 3000 series processors seem to be really efficient, concentrating heat on multiple cores in a small package could cause heat build up. We saw this with Intel when they moved over to stacked 14nm transistors with Broadwell and AMD could face something similar. Now supposedly AMD’s internal testing proves their chips do run cooler than Intel’s and results like these thermal images seem to prove that. However, we will have to test that out ourselves, especially when overclocking.
The Ryzen 3000 Family Explained
Other than the 3950X I already talked, AMD’s these lineup will have a bunch of other processors starting with the Ryzen 9 3900X, which is the first 12-core/24-thread CPU available for the AM4 platform. At $500 it is the most expensive non-ThreadRipper CPU yet, but if you want something less expensive there’s the $400 Ryzen 7 3800X and the $329 Ryzen 7 3700X. The last two processors will follow in the footsteps of the 2700X by offering 8-cores/16-threads along with higher boost frequencies. All of these have TDPs of just 105 watts, which is really impressive except the 3700X, which is 65 watts. If rumors are to be believe that will allow it to overclock quite high too.
One thing that should jump out right away is the fact that the Ryzen 7 3700X costs exactly the same amount as the 2700X when it first launched. AMD is not charging a premium for their new architecture. Part of that could be due to how similar the specs are between the 2700X and 3700X, but I really really hope Intel is paying attention here.
While the more expensive third-generation Ryzen processors will probably get a lot of attention, we are actually more excited about the 6-core/12-thread Ryzen 5 3600 and 3600X. Ryzen 5’s are usually AMD’s performance per dollar leaders and these newer ones don’t look any different.
Remember they’re competing price wise against Intel’s 6-core/6-thread i5-9600K and the i5-9400, which should be an interesting battle especially when it comes to budget gaming builds. In the above chart you can pretty clearly see why Intel is likely scrambling right now. From this perspective, there’s a lot of reason to believe that AMD will be a dominating player at every price point, but there’s some areas where AMD maybe still be a bit weak and Intel can capitalize on that. More on that in just a bit. Something else I just wanted to highlight are the two new APUs in this lineup. The 3400G and the 3200G are basically outdated Raven APUs, which still use the 14nm manufacturing process and simply have higher clock speeds versus the 2400G and 2200G. Regrettably, 7nm APUs aren’t due out this year.
Gaming = Still A Problem?
All right, so I mentioned that in some areas AMD might need improvement. If you remember with second generation Ryzen processors, they tended to lag pretty far behind Coffee Lake in single or lightly threaded workloads and because of that gaming suffered as well. Now AMD thinks that they’ve addressed that with Zen 2. When you combine their design enhancements and improved clock speed consistency, there is supposed to be a 21% increase in single-thread performance in a perfect world like Cinebench. But what about gaming? Well AMD has also doubled Ryzen’s L3 cache sizes and called it GameCache. More cache means more information can be stored locally on the processor’s internal high speed memory rather than being dumped onto the slower system memory. When you combine that with the fact that Zen 2 now supports DDR4-3600 memory speeds there are some big potential frame boosts.
But does that mean Intel should be worried? Yes and no. On the one hand AMD showed that the Ryzen 9 3900X pretty much matches the 9900K, but remember the Ryzen CPU actually costs a few bucks more than a 9900K. It’s good to see them being at an even playing field here, but the games seem cherry-picked as well to show the best in Ryzen. What’s a bit more concerning is the 3800X versus 9700K benchmarks, because they do show Intel leading in a few cases. By how much? Well there’s no way of telling, and just remember that all of these benchmarks are AMD’s and if the resolution was increased the gap would reduce by quite a bit due to the GPU being a bottleneck. Now gaming benchmarks were one of Intel’s last safe havens against Ryzen, but it looks like AMD is narrowing the gap even further with these new third generation processors.
Now when you look at it overall, there doesn’t seem to be a clear cut winner in the whole Ryzen 3000 series lineup, at least against Coffee Kake. I mean, sure they do look better, but a convincing win doesn’t look possible yet. Where Intel will lose and lose big time is in multi-threaded performance benchmarks. AMD can simply offer more cores for less money, but there’s one thing that we noticed here. There is no way AMD enabled Intel’s QuickSync support for that Adobe Premiere test, since with it on the 9900K would have dominated. Everywhere else though the wins seems pretty convincing. While virtual threads don’t scale nearly as well as physical ones, in this test we can see that AMD absolutely needs their core count advantage over Intel to win. Even though they 3800X has double the thread count over Intel’s 9700K it can only edge it out by a maximum of 37%. While that’s a huge win given the price of each CPU, there are obviously some strong points in Intel’s current architecture. Can they actually build on that? We’ll just going to have to wait and see.
Now another key selling feature of AMD’s next generation platform is the X570 chipset. As we kept digging through the information that AMD did provide us, somethings did come up that were quite concerning with regards to pricing. One of the most talked about features of X570 is the PCI 4.0 interface, which is super exciting. While graphics cards aren’t anywhere close to needing the bandwidth it provides storage devices can actually benefit. Gigabyte, Corsair, and other manufacturers have announced NVMe SSDs that can operate at 5GB per second, which is absolutely insane. A PCI 4.0 also offers a lot more flexibility for devices that need additional bandwidth. So X570 motherboards will have a lot more high-end connectivity than either Z390 or X470 motherboards. The problem we see is cost, the price of these motherboards is trending upwards. You see PCIe Gen4 traces requires upgraded PCBs for signaling strength, and many of these standard motherboards actually use workstation technologies, so that obviously drives the cost up. Another thing that’s boosting prices is the sheer number of features that many of these boards have. That’s a vote of confidence from the motherboard makers to AMD since they now feel Ryzen is the premium product everyone hoped it would become. However, finding a lower priced X570 may be a challenge.
Affordable Mobos Are MIA
To me what’s really concerning is AMD’s more affordable B500 series and A500 series motherboards are nowhere to be found. During the last two Ryzen launches the mainstream and lower-end chipsets were detailed right at the start, but this time they haven’t been mentioned at all. Now luckily AMD has almost complete backwards compatibility for a Ryzen 3000 series processors, but there are some limitations. If you want a more budget-friendly system around third gen Ryzen CPUs, maybe look for a B450 motherboard or maybe even an X470 if they go on sale. Honestly, I really think the lack of lower priced 500 series motherboards will end up hurting the Ryzen 5 3600/3600X more than Intel ever could.
Wrapping Things Up!
All right guys, so I think it’s time to bring up that question one more time. Is Intel in a spot of trouble in the desktop market? The answer to that is yes, they probably are. You see AMD’s new CPUs, shiny new CPUs, power efficiency new CPUs that are loaded with cores and they also perform very competitively against Intel and they don’t cost a lot of money. However, Intel’s biggest issue is that they don’t have a direct answer to AMD’s third generation Ryzen processors. You see the Sunny Core architecture isn’t ready for desktop yet, and even then on notebooks it’s only expected to come later in October. But then again, Intel did vaguely announce that they’re going to come out with new Core X processors, which I know for a fact are going to be super expensive. And honestly AMD already has an answer to those CPUs in the form of the Ryzen 9 3950X.
And remember it takes years to build changes into a product roadmap, and honestly Intel doesn’t have years or even months. Zen 2 and 7nm are here now, and Zen 3 with a refined 7nm process is coming in 2020. So let’s quickly talk about what Intel can do right now. Lowering their damn high prices would be a great start, but sadly they have been completely unwilling to do that in the past. First, they’ve got to start burying their faces in the sand and making believe that Ryzen doesn’t exist. AMD and NVIDIA battle it out for years with continual price cuts until their next generation cards are ready. It would work here to, Intel isn’t out of the game by a long shot, but I think they’ve stayed complacent for too long,. If the Ryzen 3000 series lives up to expectations, their entire Z390 platform looks like overpriced last gen tech.
Now of course AMD needs to actually deliver on their promises and deliver products to buy on July 7th, but judging from what we’ve seen so far I think Intel has a much bigger hill to climb. So that’s it for now guys, and I would love to know your thoughts about this whole topic. Would you consider switching to third gen Ryzen? Does the new platform look appealing to you? Let us know!