Intel will launch new 14nm processors in the second half of the year, for laptops and compact computers and will work on 10nm Coffee Lake by the end of this year.

The eighth generation of Intel has been a surprise to everyone. These processors are part of the Intel Coffee Lake family of processors, which will come out in the second half of this year, when they should have come out during the first quarter of 2018. The strangest thing is that last week it was leaked that these would be from 14nm, as are the Broadwells, Skylake and Kaby Lake. These new 14nm processors will have a new manufacturing node.

Intel with Kaby Lake has left the Tick-Tock structure, where the Tick was the arrival of a new manufacturing process and the Tock was an improvement of said manufacturing process. The new system is based on three phases, called Process, Architecture and Optimization. Intel Coffee Lake also means abandoning this process that theoretically has a few months to live. Broadwell was the Process, Skylake was the Architecture, and Kaby Lake was the Optimization. This information dismantles this principle.

According to Intel's infographic, these new processors would improve Kaby Lake's performance by 15%, but there is no data if they will be desktop or mobile processors. If we take into account the data, Coffee Lake for laptops and mobile products should come out in the middle of this year and it is possible that Intel will focus on making desktop processors in 10nm, in order to have them as soon as possible and beat AMD processors Ryzen.

It makes sense that these 14nm processors are low-power, mostly, since the Skylake-X and Kaby Lake-X should arrive in the middle of the year as an HEDT platform. Intel would have left the development of processors with 10nm architecture, for desktop processors, since AMD Ryzen would pose a problem for Intel, in theory. This means that Intel would advance Coffee Lake for desktop to the last quarter of the year, rather than hitting the market in the first quarter of 2018. We'll see what happens.

Source: arstechnica