Intel today announced availability of its 7th generation Core desktop processor family codenamed "Kaby Lake". Leading the lineup is the Core i7-7700K. Kaby Lake is the second lineup of processors for the LGA1151 socket and is launched alongside the Intel 200-series Express motherboard chipset. These chips are compatible with existing 100-series chipsets (eg: Z170, B150) provided the board has the necessary BIOS update which enables support.
In this review, we are testing the gaming performance of the Core i7-7700K, comparing it against the Core i7-6700K "Skylake" processor. Both chips will be tested across two graphics cards, the AMD Radeon RX 480 8 GB and the GeForce GTX 1080 8 GB. We also added performance numbers of the GeForce GTX 1060 6 GB after noticing surprising performance-gain differences between the GTX 1080 and RX 480 to test if our suspicion of additional CPU usage in AMD's drivers could be replicated on an NVIDIA card from a similar performance segment.
The cards were put through our recently updated batch of game tests from the VGA Bench suite. Each of the two processors is tested at three clock-speed settings - stock (out of the box speeds, untouched settings), overclock of 4.50 GHz, and overclock of 4.90 GHz on the Core i7-7700K, which the 6700K cannot achieve. We are presenting these performance numbers as percentage differences to a Core i7-7700k at stock.
The Core "Kaby Lake" family is being received with similar anticipation as Intel's time-buying 4th generation "Devil's Canyon" (Haswell-refresh) processors, which were the first signs that Intel isn't able to keep up with its "Tick-Tock" product development cadence. Quick history lesson: Back in 2008, Intel launched its game-changing Core "Nehalem" processor family on the 45 nm node, it optical-shrunk this architecture to the 32 nm node with "Westmere". The CPU core design between "Nehalem" and "Westmere" wasn't different, but "Westmere" was treated as a new generation regardless because it came on a newer process with more energy efficiency. Since then, Intel has launched new micro-architectures on "mature" processes ("tock") and new silicon-fabrication processes with "mature" micro-architectures ("tick"). This way, it wouldn't have to deal with teething problems of both the micro-architecture and the process node at the same time.
Consumers and PC makers were satisfied with new Core "generation" launches each year. The party lasted till 2014, when Intel realized it would take longer than expected to roll out the 14 nm node and optically shrink 4th gen "Haswell" to 5th gen "Broadwell." To bide time, it launched the "Devil's Canyon" (Haswell Refresh) processors on the existing 22 nm node, with higher clocks and better overclocking on the "K" variants. The company did eventually launch the 14 nm "Broadwell" as the Core i7-5775C and i5-5675C, but those weren't true successors to the i7-4790K or i5-4690K since they had a smaller cache and weren't clocked as high. It did, however, give "Broadwell" a yearlong run on the HEDT platform with the 14 nm "Broadwell-E" LGA2011v3 processors. By late-2015 or early-2016, it became clear that Intel couldn't sustain a new process node every two or so years, and that to sustain a new "generation" Core processor launch each year, it had to launch three "generations" per process node.
The 6th generation Core "Skylake" processors brought the 14 nm node to life, as they posted big performance gains over "Haswell" and were the first genuinely new chips from Intel in two years. Logically, the next generation should be "Skylake", die-shrunk to the next process node (10 nm or 7 nm), but that's not the case. We won't see the next node for over a year if reports are true. The next best thing Intel could have done is refine "Skylake" as much as it could on the existing 14 nm node. This is where "Kaby Lake" comes in. The 7th generation Core "Kaby Lake" processors feature various silicon fabrication process-level refinements that enable higher CPU core clock speeds out of the box. The company has added various new features to the integrated GPU and the platform itself, through the 200-series chipset.
The Core i7-7700K leads the pack with clock speeds of 4.20 GHz, with 4.50 GHz maximum Turbo Boost frequency. The chip further boasts of full-range base-clock overclocking. The quad-core chip features 8 MB of L3 cache. The integrated HD 630 graphics receives a host of feature-set upgrades, including support for 5K displays (at 30 Hz), 10-bit HEVC video hardware acceleration, VP9 hardware acceleration, and premium 4K Ultra HD content DRM support, which should enable the upcoming Netflix 4K option though web browsers. The 200-series chipset features four more downstream (general purpose) PCIe lanes than the 100-series chipset and supports Intel Optane (3D X-point memory SSDs).