News Intel Vies to ‘Recapture Process Leadership’ as it Returns to Two-Year Cadence

[Admin]

Intel says it wants to recapture its process lead after the plagued 10nm transition. It aims to return to a two-year process cadence, with 7nm in the fourth quarter of 2021.

Intel Vies to ‘Recapture Process Leadership’ as it Returns to Two-Year Cadence : Read more

 

[dave.jeffers]

The only thing Intel is going to "process" over the next four years is employment terminations slips, as it continues losing technology leadership, market share, and money to AMD.

 

[NightHawkRMX]

Intel isn't going anywhere. Sure, AMD is gaining on them, but Intel has far higher market share in the server market they have a lot of time to get a good product out.

 

[JayNor]

wikichip has a write-up on that 10nm NNP-I/Spring Hill chip, including info that it has 8.5B transistors and 239 mm² die size. That puts it at about twice the area of the ice lake chip, and it has a power envelope up to 50W. There is no indication that this is a High Performance 10nm process, but it is interesting to see that they have some performance available in that HD process that they have been holding back from the laptop chips.

https://en.wikichip.org/wiki/intel/microarchitectures/spring_hill#SoC_Overview

 

[DotNetMaster777]

Nice that lost in past is found and intel is back without any issues !!

 

[cryoburner]

With Intel squarely stating it wants to go back to a two year cadence, that implies that Moore’s Law will live on for a few nodes longer.

I'm not sure it counts as being on a "two year cadence" when the only reason their 7nm process is supposedly coming relatively soon after 10nm is because 10nm was delayed for so many years. Intel was similarly talking about how they were going to maintain a process advantage with 10nm around the time they launched 14nm back in 2014. Then, 10nm was delayed until late 2017, then 2019, and we still have no 10nm desktop processors even on the roadmap for 2020. It will have taken at least around 6 years for Intel's desktop chips to see a process node shrink from 14nm by the time they come out.

Also, these nodes are not as well-defined as they used to be, with only certain parts of the processor dropping in size. Their original 10nm design was apparently not tenable, so they scaled it back to get what they are now calling 10nm, and a few years later than expected, at that. Their 7nm process might end up being more comparable to what they had originally planned for 10nm to do a few years ago.

I would not consider that as "Moore's Law living on". Process advancement has clearly slowed, and we are not seeing the number of transisters in a given area doubling every two years, by any means.

 

[JayNor]

Intel did mention 5nm in the cc.

I think, though, that Intel is planning to advance thelr 3D manufacturing lead. Keller has a plan for 50x gate density

https://www.pcgamesn.com/intel/jim-keller-building-nanowires-intel-tsmc-samsung

 

[InvalidError]

The only thing Intel is going to "process" over the next four years is employment terminations slips, as it continues losing technology leadership, market share, and money to AMD.

AMD has zero market share in networking, FPGA, AI-specific accelerators and a handful of other things. Corporate buyers take forever to switch vendors, so most of those sales aren't going anywhere any time soon either. Since Intel's revenue went up 10% despite 8% lower client CPU sales, Intel is clearly more than making up for lost sales in the consumer space elsewhere, not in any meaningful danger.