Over the past few weeks, chatter has increased about Rocket Lake’s performance and corresponding benchmark leaks. We’ve seen a number of Geekbench results for various CPUs and written a few comparisons ourselves. Since the leaks keep coming, it seemed more helpful to round up expectations for RKL and how it’ll likely compare to Zen 3 / Ryzen 5000, rather than continue dissecting each leak individually. We’re comparing CPUs here rather than platforms, but Rocket Lake also introduces features like PCIe 4.0 for Intel CPUs.
Rocket Lake uses Intel’s Cypress Cove CPU architecture. Cypress Cove is a backport of Intel’s 2019 Ice Lake architecture, built on the company’s older 14nm process. Intel has continued using 14nm for desktops, despite having 10nm in-market for mobile, because the characteristics of the node have been better suited to building high-performance multi-core CPUs. Cypress Cove is expected to offer ~1.19x higher IPCs than Comet Lake, and it brings AVX-512 support to consumer systems. But how will this actually compare against AMD?
First, we expect Rocket Lake to deliver strong single-thread performance improvements. Ever since AMD launched Ryzen, Intel has emphasized high single-threaded performance. AMD claimed the gaming market with the launch of the Ryzen 5000 series, but Rocket Lake offers Intel a chance to seize it back again. Various leaked results suggest Rocket Lake’s single-threaded performance is somewhere between 1.17x – 1.35x faster than the equivalent Comet Lake processor. Either would represent respectable generational uplift.
Intel will be at its strongest against AMD in moderately threaded applications that don’t scale well above eight cores. The expected IPC gain is high enough to give Intel a real shot at recapturing gaming overall.
Rocket Lake’s multi-threaded performance is a little harder to predict. The same leaks that suggest significant single-threaded leaps sometimes predict smaller multi-threaded improvements of 1.1x – 1.25x depending on the application and the CPU. Some leaked results use engineering samples, not retail chips, and a wide set of results is not unusual. The most positive leaks suggest that the Core i9-11900K will match or slightly exceed the Core i9-10900K’s performance, even though it only fields eight cores to its predecessors’ 10. The less flattering predictions suggest the Core i9-11900K will come in just behind the 10900K. We’re likely to see both these results in practice, depending on the degree to which an application is or isn’t fond of Intel’s Cypress Cove architecture.
We don’t expect dramatic improvements in power consumption, since Intel is still working with the same 14nm node. We expect any power efficiency improvements Rocket Lake delivers to arrive courtesy of improved IPC as opposed to intrinsically lower transistor-level power consumption. If Intel has managed to improve IPC and absolute power consumption, so much the better.
Rocket Lake isn’t going to beat AMD’s entire product stack. CPUs like the 5900X and 5950X will remain absolute performance leaders in well-threaded applications. But Intel isn’t trying to demonstrate that its eight-core chip can beat a 16-core CPU. Rocket Lake is intended to show that Intel can still move the desktop forward and improve IPC, while hopefully winning back specific segments like gaming. Back in 2018, Intel told the assembled press it was working on a new process-agnostic design methodology that would allow it to move CPUs between nodes much more easily than before. Rocket Lake is a step towards that flexibility.
Overall, we expect Rocket Lake to seriously challenge AMD on single-threaded performance, particularly in gaming. In multi-threaded, we expect the Core i9-11900K to match or slightly exceed the 5800X. AMD’s Ryzen 9 5900X and Ryzen 9 5950X will outpace Rocket Lake in any multi-threaded performance comparison, but that’s to be expected. Performance per watt is likely to favor AMD. Performance per dollar will depend on how Intel prices its CPUs. If Intel prices aggressively, it might be able to pull off an upset against AMD at several price points. If it sticks to its typical behavior and prices well above an equivalent AMD CPU, Team Red will probably continue to retain an overall price/performance advantage.
Why Apple’s M1 Chip Threatens Intel and AMD
Intel's own history suggests it and AMD should take Apple's new M1 SoC very seriously.
How L1 and L2 CPU Caches Work, and Why They’re an Essential Part of Modern Chips
Ever been curious how L1 and L2 cache work? We're glad you asked. Here, we deep dive into the structure and nature of one of computing's most fundamental designs and innovations.
How to Be a Task Manager Wizard, According to the Guy Who Wrote It
The author of the original Windows Task Manager has some tips on how to use it more effectively, including a few we'd never heard of — and one we tossed in ourselves if you have trouble with Task Manager getting stuck behind fullscreen-stealing windows.
What Does It Mean for the PC Market If Apple Makes the Fastest CPU?
Apple's M1 SoC could have a profound impact on the PC market. After 25 years, x86 may no longer be the highest-performing CPU architecture you can practically buy.