It is difficult to say how superstitious the current head of Intel is, but his regular promises to return the rhythm of the change of technological processes on the company’s conveyor by two or two and a half years will not come true. At the quarterly reporting conference, he had to declare that the development of 7-nm technology is delayed for a period of six to twelve months.
Image source: Intel
–
–
–
Tellingly, Intel management shares the concept of “product delay” and “lag in the schedule of reducing the defect rate.” If in the latter case Intel is 12 months behind its own plans due to a defect found in 7nm technology, it intends to delay the release of 7nm products by only six months. How is this possible? Everything can be explained simply – Intel now does not hesitate to more actively use the production facilities of third-party companies. The names of the latter are not disclosed, but TSMC, the largest contract manufacturer in the world, comes to mind first.
As our regular readers may remember, earlier Intel promised to present the first 7nm product in the face of the Ponte Vecchio computing accelerator at the end of 2021, and only in 2022 it will launch the release of central 7nm server processors, which received the symbol Granite Rapids. At today’s quarterly reporting conference, CEO Robert Swan confessedthat the Ponte Vecchio accelerator was originally conceived as a kind of “debug” product, following the example of Lakefield, since with its help Intel hoped to gain experience in creating products with complex spatial layout.
Image source: Intel
–
–
–
Swan explained that the Ponte Vecchio should include various stand-alone crystals, and Intel originally planned to produce some of them on the side. Now, when obvious problems arose with the development of the 7-nm technical process on their own, the question arose about transferring the production of one of “our” crystals to the contractor’s conveyor. Apparently, it is the announcement of Ponte Vecchio that will be delayed by six months, although at the last conference the Intel leadership only vaguely designated the period from late 2021 to early 2022.
In general, the term “die disaggregation” was most often sounded from the lips of Robert Swan. Multichip layout will become the norm for the vast majority of future Intel products, it will allow some of the crystals to be outsourced. Swan promises that Intel will take a pragmatic approach to the distribution of orders for the production of certain components. If outsourcing in a particular case will speed up the release of final products to the market, then the company will use it without undue hesitation.
The timing of the appearance of the first 7nm client processors was announced – this should happen in late 2022 or early 2023. The layout and designation were not mentioned, but it can be assumed that we are talking about processors of the Meteor Lake family. In the first half of 2023, deliveries of 7nm processors for server applications will begin. They were previously thought to be released in the first half of 2022. In other words, we can talk about a shift in terms of one year. This is natural, since Intel will certainly continue to produce central processors on its own, and in this case it does not have much room for maneuver with the involvement of third-party manufacturers.
Intel also promises to update its processor line every year, even if it happens through the same lithographic process. By the end of 2022, such an update will be made in the customer segment. Considering that the debut of 10nm Alder Lake processors is promised in the second half of 2021, then we can talk about their improved version.
–
If you notice an error, select it with the mouse and press CTRL + ENTER.
/data/photo/2018/09/06/4127596084.jpg)