Intel is gearing up to roll out the first non-Skylake based architecture in what is the better part of a decade and leaks have started popping up left and right. The new branding of Intel's upcoming 500 series chipset has leaked out courtesy of Videocardz and Rocket Lake's flagship processor has broken through the 700 single core score barrier on CPUZ. It also defeated AMD's Zen 3 in benchmarks we covered a few days ago.
Intel Core i9-11900K Rocket Lake CPU sets new CPUZ single core record, Z590, B560 and H510 chipset branding revealed.
The new branding is as follows:
The following screenshot leaked out on Chiphell and shows a new score set by Intel's Rocket Lake flagship. This also the first time that the CPU has breached the 700 point mark. The Intel Core i9-11900K achieves 706 points on the single core side which is the strongest single core score in x86 processors. This also means that the company has taken back the single thread lead that it enjoyed for so many years.
Twitter user @harukaze5719 also compiled the benchmarks for ease of comparison and you can see how it compares to AMD's latest Zen 3 architecture:
Intel's Rocket Lake processors will have up to 8 cores and feature 50% more integrated graphics power. RKL-S is based on the 14nm process, but unlike older generations, it will feature a revamped architecture based on the Sunny Cove (Ice Lake) architecture called Cypress Cove. It is essentially a backport of 10nm (non-SuperFin) to 14nm. PCIe 4.0 support and Xe 12 generation graphics have been added as well.
Since it is based on the 14nm process, leaked benchmarks indicates that it will be able to boost up to 5.0 GHz (which is typical for the highly mature process right now) allowing for a best of both worlds situation where a new architecture is able to hit the ground running (speaking from a clock speed perspective) because of the highly mature node.
Rocket Lake S will feature 20 CPU based PCIe 4.0 lanes (4 more than last generation) making it the first mainstream Intel architecture to support the PCIe 4.0 standard and something that has been sorely lacking from the company's lineup right now. It features new Deep Learning Boost and VNNI support which should give a significant speedup to AI-related workflows. It also features the new USB 3.2 Gen 2x2 standard for blazing-fast USB speeds and has Intel's 12th generation Xe graphics iGPU - which is nothing to sneeze at. Xe Graphics introduces support for high end video decoders like 4:4:4 HEVC and VP9 while allowing display resolutions of up to 3x 4k60. The integrated memory controller now natively supports speeds of up to DDR4-3200 without any overclocking.
This is great news for enthusiasts because RKL is going to be the first major new architecture in a very long time. However, the fact that it is based on Cypress Cove instead of Willow Cove likely means that we will not see the 25%+ IPC we were expecting although Intel has already confirmed that it *will* be double digits. A 5.0 GHz boost clock combined with a major new architecture should offer the best possible combination for a performance upgrade for Intel customers (although at the cost of similar TDP) in a very long time and I am personally really excited to see how this performs.