According to leaked documents about Intel's upcoming plans in the processor market, the company's future does not appear very promising. If the documents are accurate, the increase in the number of cores in mainstream processors to ten will not happen until at least 2020, with 14nm processors dominating the desktop segment until 2022. The much-debated 'thin' 10nm process will be exclusively tested in the mobile segment on energy-efficient U- and Y-series processors. However, trial shipments of Ice Lake could begin as early as mid-year, but widespread availability of mobile 10nm chips will also have to wait—at least until mid-2020.
Intel's 'roadmap' with such surprising revelations was made available to journalists from the Dutch site Tweakers.net. The publication indicates that the source of the slides on the plans is a presentation from one of the leading partners of the microprocessor giant, Dell. However, the relevance of the presented materials remains in question, although all past announcements are accurately described within.
According to the provided data, the next update of mainstream processors for desktop systems is planned only for the second quarter of 2020, when Coffee Lake Refresh will be replaced by processors with the codename Comet Lake. At the same time, the information that Comet Lake may feature modifications with an increased number of computing cores up to ten is confirmed. Nevertheless, for the production of Comet Lake, the microprocessor giant will continue to use the 14nm process. Furthermore, the next generation of CPUs for the desktop segment, expected in 2021, is also not planned to switch to a more advanced process technology and new microarchitecture. The anticipated Rocket Lake processors will continue to be produced using 14nm technology, still offering no more than ten computing cores.
It can be concluded that desktop users will gain access to Intel processors manufactured using more modern processes only in 2022. These are likely to include some solutions based on 7nm technology, featuring a progressive microarchitecture of the Cove class, such as Golden Cove or Ocean Cove. However, the existing stagnation is expected to continue for the next two and a half years. It's worth noting that at the beginning of 2021, Intel plans to update the platform by introducing support for PCI Express 4.0. At least, this intention has been indicated for entry-level Xeon E processors, which traditionally share the same semiconductor base as consumer Core processors.
As for the mobile segment, surprisingly, the microprocessor giant plans to introduce 10-core 14nm Comet Lake processors there as well. However, it is clear that these will be niche solutions with a thermal package exceeding 65 watts. The more suitable Comet Lake U-series processors for thin and light systems will have a TDP of up to 28W and will feature up to six computing cores, while the Comet Lake Y-series with a TDP around 5W will have either two or four cores. The arrival of the Comet Lake design in the mobile segment is expected to coincide with the desktops in the second quarter of 2020.
The widespread adoption of mobile processors produced using 10nm technology is expected only at the beginning of 2021. That is when Intel plans to start manufacturing four-core Tiger Lake U and Y series processors with four computing cores and the new Willow Cove microarchitecture. However, as a backup, Intel also plans to release mobile 14nm Tiger Lake processors at the same time, indicating some uncertainty in the company regarding its own capabilities.
However, Intel must still fulfill its previously made promises that systems built on 10nm processors will be available in stores by the end of this year. The announcement of the 10nm Ice Lake processor, featuring two and four cores and the fundamentally new Sunny Cove microarchitecture, is scheduled for the second quarter of this year (likely to occur during the ComputeX exhibition). However, an important note in the documents states 'limited', indicating that Ice Lake supplies will be limited. It's hard to say what this means, especially considering that Intel has been providing limited supplies of 10nm processors for a whole year—specifically, the dual-core Cannon Lake without a graphics core.
The company's plans also specifically highlight the upcoming announcement in the second quarter of this year for the Lakefield processors—comprised using Forveros technology in multi-chip systems-on-chip with a TDP of 3–5 watts, which will simultaneously feature one 'big' 10nm Sunny Cove core and four 10nm Atom-class cores. It is worth noting that such solutions are designed by Intel for specific clients, so they won't be mass-produced either.
Therefore, if the published information about Intel's plans is accurate, we should prepare for the fact that the company's problems arising from the unsuccessful transition to the 10nm process will not disappear in the near future. The echoes of these issues will continue to follow the microprocessor giant until 2022, and they will particularly impact the desktop segment.
Source: 3dnews.ru
