Old vs new: Why the Raptor Lake refresh might be more competitive than you think

Aug 22, 2023

While we'd love to see new Intel CPUs on the desktop, what we're getting with 14th-generation Raptor Lake chips isn't bad either.

Until somewhat recently, we expected Intel would continue its comeback story with Meteor Lake, manufactured on the formerly troubled Intel 4 process (previously called 7nm). However, one or two generations of good but late 10nm CPUs like Alder Lake couldn't undo five years of 14nm Skylake refreshes, and some of Intel's worst CPUs on its own, and AMD certainly isn't slowing down. After the 13th-generation Raptor Lake in 2022, it was natural to think it would be followed up with Meteor Lake sometime in 2024, maybe 2023, if we're lucky.

However, it's all but certain that Meteor Lake is destined to become a laptop-exclusive processor, even though, on paper, it would have been capable enough for the desktop. All signs point to these 14th-generation desktop chips being refreshed Raptor Lake chips, and while Intel hasn't outright said one way or the other, the company's upcoming branding changes practically confirm it.

While many are understandably unhappy that Intel is refreshing a CPU that was already sort of a refresh to begin with, I don't think this will be another 14nm situation. In fact, I think 14th-generation chips will compete better than you might expect against current-generation Ryzen 7000 CPUs and future Ryzen 8000 chips.

While Intel had planned for years to follow up 10nm desktop CPUs with Meteor Lake, those plans were dropped, perhaps as recently as a year or two ago. It's a little early to pinpoint exactly why Intel axed its Meteor Lake desktop CPUs and give up an advantage over AMD, I think the thing that changed Intel's mind the most was Raptor Lake itself.

While Alder Lake and Meteor Lake were rumored as early as 2018, Raptor Lake only popped up in late 2020, which strongly hints that Intel originally intended to move on from Alder Lake and immediately onto Meteor Lake for desktops. As we know today, Meteor Lake still hasn't launched. A two or even three-year gap between CPU generations would be unthinkable for Intel, and the company would have seen this coming by 2020, hence a new 10nm CPU.

Ultimately, it seems Intel has decided on a Raptor Lake refresh in order to retain its existing price structure.

But even beyond the need to fill the gap between Alder Lake and Meteor Lake, Raptor Lake also needed to exist in order to keep Intel competitive since AMD would launch new CPUs in 2022. So Intel took Alder Lake, doubled the E-core count, added some L2 and L3 cache, and took advantage of 10nm's relative maturity to increase frequency. The end result was that Intel's 13th-generation CPUs have become some of the best CPUs you can buy today.

From here, the choice for Intel was between Meteor Lake and a Raptor Lake refresh. If Intel went with Meteor Lake, it would have to raise prices or just accept less profit to account for higher production costs associated with the new 7nm node and the tile system. While Meteor Lake desktop CPUs wouldn't have been slower than Raptor Lake, they probably wouldn't have been much faster since they lack two P-cores. Ultimately, it seems Intel has decided on a Raptor Lake refresh in order to retain its existing price structure.

So, what can we expect out of the Raptor Lake refresh? More cores at the same prices it seems, just like with the original 13th-generation Raptor Lake CPUs. It's still not clear what the product stack will look like, but rumor has it that Core i3 chips will start at six P-cores rather than four, and that the Core i5-14600K and Core i7-14700K will have more cores than their predecessors. High power consumption will still be a problem, but a core count increase all across the board would be great even still.

Products always have to be evaluated against the competition; even if a new product is better than the one it's replacing, why would you care if some other company's product is the best anyway? For AMD, this might seem like a pretty good matchup since it's going up against the same CPUs again, and this time it'll have newer chips. Ryzen 7000 already performs favorably against Raptor Lake in efficiency and high-framerate gaming, so this should be a walk in the park, right?

Well, firstly, time is not on AMD's side here. The Raptor Lake refresh is practically guaranteed to launch by the end of 2023, while Ryzen 8000 is slated for 2024, and will likely hit later in the year as AMD hasn't disclosed much about the new chips and the rumor mill hasn't churned out any credible tidbits. And it'll probably cause headaches for AMD initially as Ryzen 7000 already struggles against Raptor Lake due to high costs for the AM5 platform and a lack of low-end Ryzen 7000 CPUs.

Intel could very well pull off a hit-and-run attack with Raptor Lake if Arrow Lake launches just in the nick of time.

I don't see Ryzen 8000 making a ton of progress against the Raptor Lake refresh either, even though, on paper, it seems quite good based on what we know. AMD has disclosed Zen 5 chips will be built on TSMC's 4nm and 3nm process, but we have every reason to believe Ryzen 8000 will use the newer 3nm process since it will almost certainly utilize the same chiplets data center Epyc CPUs do.

However, while 3nm is the latest, that doesn't mean it's the greatest. TSMC's 3nm offers either 30% less power consumption or a 15% frequency boost (or a mix of the two) compared to 5nm (which Ryzen 7000 uses), which isn't bad a bad upgrade, but its density improvement is one of the worst we've ever seen. In TSMC's mainstream 3nm process, logic transistor density (which matters for cores) only increases by 70%, but that's amazing compared to the 0% gain in SRAM or cache density. The company will offer a slightly different 3nm with a 5% denser cache, but AMD probably won't use this version, and 5% wouldn't even make a big enough difference.

By contrast, the jump from 7nm with Ryzen 5000 to 5nm with Ryzen 7000 offered a 40% lower power draw or a 20% clock speed increase (AMD took more of the clock speed gain with Ryzen 7000) plus an 80% increase in logic density and a 35% increase in cache density. TSMC's 3nm node makes it so much harder for Ryzen 8000 to compete with the Raptor Lake refresh, not just because the performance gains won't be as good as with 5nm, but also because newer nodes have higher production costs than older ones, so don't expect Ryzen 8000 to spawn any great budget CPUs.

And this is all assuming AMD can even get Ryzen 8000 out fast enough to face off against the Raptor Lake refresh rather than the upcoming Arrow Lake chips, which are rumored to offer the same 8 P-cores and 16 E-cores as Raptor Lake while using the newer 20A node. Intel could very well pull off a hit-and-run attack with Raptor Lake if Arrow Lake launches just in the nick of time. Even if Ryzen 8000 launches several months or almost a year before Arrow Lake, it's not like increased production costs will make it any easier for AMD to compete in the budget segment without relying on old Ryzen 7000 chips.

Intel is content to iterate on older silicon until it's ready to launch the next big thing. AMD is always looking for the most advanced ways to manufacture its CPUs to create cutting-edge products for each generation. Where Intel can offer highly competitive and modern CPUs below the $200 mark, AMD can consistently provide high-end CPUs for gaming and content creation with the best efficiency and highest performance. The next generation or two will really highlight how differently Intel and AMD approach desktop CPUs.

The Raptor Lake refresh won't really move the needle for performance, but it definitely will give us CPUs with even better value, something that's hard to imagine AMD ever offering. It's a shame we won't be able to enjoy Intel's first chiplet-based CPU on the desktop for another year or two, but sometimes that's how it goes. As long as Intel gets Arrow Lake out on time (and this is Intel we're talking about, so knock on wood), then I think we can forgive Intel for launching just one more refresh.

Matthew is a PC Hardware Writer at XDA, having previously written for Digital Trends, Tom's Hardware, and other publications since 2018. He's mainly interested in the three way fight between AMD, Intel, and Nvidia in PCs and servers. He also has a degree in history and linguistics.