For a long time, Intel has applied a “tick-tock” model of product development. According to it, new manufacturing technology was first introduced to an existing architecture (tick); then, using the same process technology, Intel brought out a new architecture (tock). Increasing costs for smaller manufacturing techniques and the laws of physics put a stop to this successful model.
To solve this problem Intel created what they call the Process-Architecture-Optimization. Process-Architecture-Optimization is Intel’s new development model introduced as a replacement for the phased-out Tick-Tock model. Where the Tick-Tock model used two generations of processors with a single manufacturing process the new model will use three. The objective now is to take the existing architecture and introduce it on a new manufacturing technology (processing), then replace it with a new architecture (architecture) which then can be optimized (optimization), before it is once again time for a new and smaller manufacturing technology (processing).
Kaby Lake is step three under the new development model (optimization), introduced in the fall 2016 for portable computers. In the beginning of 2017 Kaby Lake made its entrance for the full line of processors. Optimizations are the same as for the portable market, meaning that not much has changed from the previous generation.
Kaby Lake is basically about the exact same architecture as Skylake, where features like the number of cores and the performance at a given clock rate (IPC) remain unchanged. The graphics portion is unchanged performance-wise; Intel has however updated the media features to provide better support for higher resolutions. Where Skylake accelerated 1080p HEVC encoding, Kaby Lake now does native hardware acceleration of 4K HEVC encoding/decoding at 10-bit depths and VP9 decoding.
Regardless of nonexistent architectural improvements, Kaby Lake still presents a slight performance boost compared with Skylake. Kaby Lake is based on a refined variant of Intel's over two-years-old 14 nanometer technology, 14nm+, which will provide about 12 percent better performance (clock frequency) than Skylake without increased power consumption.
Another improvement with Kaby Lake is the second generation of Speed Shift, a technology that first saw the light of day with Skylake. With Kaby Lake, the hardware control around Speed Shift has improved and the CPU can now reach peak frequency in 10-15 milliseconds instead of around 30 milliseconds with Skylake, something that will result in a noticeably more responsive experience in everyday life.