Slightly ahead of the rumoured schedule, Intel has released their latest update to their CPU lineup. Dubbed Kaby Lake, the new chips feature tweaked versions of the previous generation 14nm process and Skylake architecture. Somewhat disappointingly, only mobile chips will launch this month while we will have to wait another quarter before the desktop chips arrive next year.
As the first “Optimization” after the new Process-Optimization-Architecture strategy replaced Tick-Tock, Kaby Lake largely fulfills expectations. As part of the optimizations, Intel isn’t claiming any real IPC improvements. Instead, the stated performance gains mostly come from higher stock clocks and the improved SpeedShift which allows even faster response to dynamically boost clocks. The biggest feature appears to be 10-bit HEVC/H.265 which is welcome but hardly groundbreaking.
The other big optimization is the use of the new 14nm+ process. This brings improved fin profile and improved transistor channel strain which reportedly brings about 12% performance improvement. How this plays out in terms of efficiency and peak clock speeds remains to be seen. intel is being a bit light on details so we’ll hopefully get more information as the chips continue to roll out.
For now. we’re getting six chips spread over two different lineups. First up we have the new Core-Y which replaces the Broadwell Core-M lineup. Changing it to Core-Y makes a bit of sense as the next lineup is the usual U series of CPUs. Both are low powered mobile chips with Core-Y coming in at 4.5W and the U-series at a higher 15W TDP. Just like previous low-powered chips, we’re only getting the 2+2 units, or 2 CPU cores with hyperthreading and 2 GPU clusters. Both feature the integrated chipset as well, supporting similar configurations for Skylake to enable drop in replacement.
With all this said and done, it looks like Kaby Lake won’t be bringing any major improvements over Skylake. Of course, for the overclocks among you, we will have to wait until 2017 to see where the new 14nm+ process will take us. Hopefully, future optimizations provide a bit more of a boost.
