Finally, new information regarding to the new architecture. An update on Git Clang LLVM compiler from Intel reveals support for the company’s 8th generation Core “Cannonlake” processors, the successor to the successful chip 7th Generation Kaby Lake (the new Skylake CPUs) which come out later this year and is targeted at a 2017 release. New data confirm that Cannonlake will introduce some new instruction sets such as AVX-512 (abx512ifma and avx512vbmi), new extensions SHA, and PICU.
With the IFMA and VBMI extensions, the AVX-512 instruction sets on Cannonlake, the new series ranks as the largest CPU in terms of extensions that can be found in any CPU of the company, for example Skylake is limited to avx512f, avx512cd, avx512dq, avx512bw and avx512vl. I believe you remember that Cannonlake has been delayed for several times, the latest fixed date is “mid-2017 “. The reason for the delay is unknown, it may be due to the problem in the transition to a node or even smaller than we already mentioned. Well, at the moment there is no competition in the market, so Intel isn’t in a hurry and it making sense to return on the existing new 14nm CPUs as long as possible. Intel Cannonlake will be company’s first CPUs to manufacture on 10nm process node.