AMD has stolen the headlines with its Zen 3-powered Ryzen 5000 series processors that will come to market on November 5, but not to be left out of the hype, Intel is now sharing more details about its competing Rocket Lake processors that will arrive in the first quarter of 2021.
Intel’s Rocket Lake processors are among the industry’s worst-kept secrets, with plenty of information already having come to the surface. Intel’s new press release verifies much of what we already knew, such as Rocket Lake’s planned double-digit increase in instruction per cycle (IPC) throughput. That’s a much-needed advance in the face of AMD’s Ryzen 5000 processors, which feature a 19% increase in IPC. The increased IPC benefits all aspects of performance and power consumption for both Intel and AMD, impacting everything from professional applications to gaming.
Intel also bumped the memory clock speeds up to DDR4-3200, a nice increase over the current DDR4-2933 supported with Comet Lake processors. That increased data transfer rate may play a role in the higher IPC, too.
Intel also confirmed that Rocket Lake tops out at eight cores and 16 threads, which is a step back in core counts from Intel’s current Comet Lake processors that top out at ten cores. That means the company will lean on the improved IPC and possibly clock speed enhancements to take on AMD’s core-heavy Ryzen processors that top out at 16 cores for the mainstream desktop. Intel also lists a 125W TDP and (250W under boost) for the chip it used to provide its vague performance predictions.
Intel finally confirmed one tidbit that wasn’t entirely clear: The new chips feature the Cypress Cove architecture, Intel’s first new microarchitecture for the desktop PC since Skylake arrived back in 2015. Intel says this new architecture is based on Ice Lake’s ‘Sunny Cove’ architecture and also comes with the same performant 12th-gen Intel Xe LP graphics engine found in the Tiger Lake processors.
It’s notable that while the chips come with a new microarchitecture, they still purportedly leverage Intel’s 14nm process (Intel didn’t clarify in the materials). We know that the Cypress Cove microarchitecture hasn’t made an appearance in Intel’s roadmaps, so the branding is likely meant to indicate that the design is a back-ported version of the Sunny Cove cores found in Ice Lake processors. “Back-porting” is a method that allows Intel to take a new design built on a smaller process node, in this case 10nm, and etch it on an older, larger node (in this case 14nm).
Rocket Lake also marks Intel’s first desktop PC chips that support the PCIe 4.0 interface, a needed addition that comes two long years after AMD led the industry with the first PC chips to support the interface. Intel also reworked the internal PCIe subsystem to accommodate a x4 direct connection (the chips now support 20 lanes), with the ‘extra’ four lanes being rumored to be a direct connection from the CPU to NVMe storage.
- New Cypress Cove architecture featuring Ice Lake Core architecture and Tiger Lake Graphics architecture.
- Double-digit percentage IPC performance improvement
- Better gen-on-gen performance
- Up to 20 CPU Pcie 4.0 lanes
- Intel UHD graphics featuring Intel Xe Graphics architecture
- Intel Quick Sync video, offering better video transcoding and hardware acceleration for the latest codecs
- New overclocking features for more flexible tuning performance
- Intel Deep Learning Boost and VNNI support
The chips drop into 500-series motherboards, and we expect them to work with existing 400-series motherboards as well. However, we can expect a new wave of 500-series motherboards to arrive at the launch.
Intel also touted support for its Deep Learning Boost (DLBoost) and VNNI features, which require support for AVX-512 instructions. Thus, Intel has confirmed support for AVX-512 with Rocket Lake, and both DLBoost and VNNI will enhance performance in workloads that leverage AI algorithms. Intel has been fostering the software ecosystem to better take advantage of the explosive performance benefits. Adobe’s recent suite of updates is a prime example of the improvements we can expect as AI becomes more broadly adopted for PC workloads.
Enthusiasts can also look forward to new undefined overclocking features, which would be helpful as Intel’s overclocking headroom, while shrinking, still tends to be a big advantage over AMD’s chips.
Intel also added new features to the graphics engine, with new HEVC, VP9, and SCC encoders that support up to 4K60 12b and hardware-accelerated AV1 support for 4K60 10b. Keep your eye on AV1; this seems to be the looming industry standard. Intel also touts the increased display resolution, with now stretches up to three screens at 4K60 or two screens at 5K60.
Intel says the chips will come to market in the first quarter of 2021, which means we’ll likely see them sometime in April of next year, ceding the lucrative holiday shopping season to AMD’s Ryzen 5000 processors.