[ad_1]
The reports begin to suggest that Intel is ending its efforts in the 10 nanometer manufacturing field and all are evoking an article published today on SemiAccurate. According to the article, "Intel has just put an end to its 10 nm process in trouble", which is a "good thing to do". Whether or not the right choice is irrelevant – Intel says the reports are wrong.
I contacted Intel about this and it was told to a white point that the reports are "fake". Intel also referred me to a Twitter article saying essentially the same thing, adding that it "is making good progress on 10 nm" and that "yields are improving".
The media reports published today that Intel is terminating work on the 10 nm process are false. We are progressing well on 10nm. Yields improve according to the schedule we shared in our last earnings report.October 22, 2018
The fact that yields are "consistent with the timeline" recently shared by Intel means that Cannon Lake is still about to ship in volume during "2019 vacation." It's the last official word we've heard from Intel, anyway.
Reaching 10 nm has not been without difficulties. Matrix reductions are tricky at this level, and Intel would say (and do) that its manufacturing process is more advanced than that of the competition. Mark Bohr of Intel explained this in a blog post last year, claiming that some companies "continued to advance node names even in cases where the density increase was minimal or zero ".
"The industry needs a standardized density measurement to standardize the rules of the game. Customers must be able to easily compare process offers from a chip maker and those from different chip makers. The challenge lies in the increasing complexity of semiconductor processes and variety of designs, "said Mohr.
It's worth reading because it talks about grid pitch, logical transistor density, SRAM cell size, and other technical details in a short blog post.
At the same time, 10nm did not arrive as fast as we hoped, at least in volume. The only Intel 10nm chip is a mobile processor without a built-in graphics card, and its availability is comparatively limited.
It would not shock us if Intel once again delayed its 10-nm mass deployment, but as it stands, Cannon Lake is yet to arrive at the end of next year.
[ad_2]
Source link
