Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/intel-delays-mass-production-of-10nm-cpus-to-2019.10363/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Intel delays mass production of 10nm CPUs to 2019

“We are shipping [10-nm chips] in low volume and yields are improving, but the rate of improvement is slower than we anticipated. As a result, volume production is moving from the second half of 2018 into 2019. We understand the yield issues and have defined improvements for them, but they will take time to implement and qualify.”

“We are going to start that ramp as soon as we think the yields are in line, so I said 2019,” Mr. Krzanich noted. “We did not say first or second half, but we will do it as quickly as we can, based on the yield.”

The Intel 2 year process tick-tock stopped with 14nm being 3 years and 10nm 4 years. In my opinion Intel got caught up in the transistor race with TSMC and finally crashed.

For me this all started when Intel decided to get into the foundry business again then took Altera, TSMC's close partner. I was at the Starbucks in TSMC Fab 12 when it was announced. It really did make a big impact on TSMC employees, shock and disbelief.

Do you remember when Mark Bohr from Intel said the fabless model was collapsing? That the foundries would not be able to follow Intel where they are going?

Intel says fabless model collapsing... really?

More than 88k people read that blog by the way...

Bottom line: Social media sites like SemiWiki and Anandtech brought transparency to the semiconductor industry and transparency has brought embarrassment to Intel and that embarrassment continues. Just my opinion of course.
 
Last edited:
They won't. TSMC is ramping up 7nm and will be at 5nm in 2020. Intel will be 2-3 years behind and will be at a major process disadvantage. No wonder Apple planning to switch, and everyone else will follow.
 
Intel still makes the best product. Yes there's a niche market for amd but for most of us intel is the best. I think the problem is that you can only go so far banging your head against the wall and let's be honest that's how most engineering is done. Everyone's struggling not just intel.

You can criticize Mark Bohr as being an empty suit but he probably thinks he got paid and really doesn't care. There are a lot of people like that.
 
Portland said:
You can criticize Mark Bohr as being an empty suit but he probably thinks he got paid and really doesn't care. There are a lot of people like that.
Click to expand...

I met Mark at a conference and he is definitely not an empty suit. The problem is that Mark has never been exposed to the type of transparency that we have today. The same can be said for politicians and other public figures. After 30 years inside the Intel closed loop culture it must have been quite an awakening for Mark and Intel on a whole. Now marketing hype can only go so far before you are destroyed by the transparency police.
 
Daniel Nenni said:
I met Mark at a conference and he is definitely not an empty suit. The problem is that Mark has never been exposed to the type of transparency that we have today. The same can be said for politicians and other public figures. After 30 years inside the Intel closed loop culture it must have been quite an awakening for Mark and Intel on a whole. Now marketing hype can only go so far before you are destroyed by the transparency police.
Click to expand...

I'm playing devil's advocate but Intel is making record profits so everything is going well.

That said I've seen companies with record profits lose their mechanic and the company fails because no one knows how to run the automated equipment. Oregon is a very cliquey and catty. The same problems the campus in Washington County is having happens at the college level, other companies, with neighbor and really all over. Israelies they stick to a philosophy and do it. In Oregon politics come first.
 
Intel's record profits are an act of financial engineering. You change the depreciation schedule of a Fab from 3 years to 5 years and boom, depreciation expense goes down 40% and profits magically go up. You acquire a bunch of companies with debt and revenue goes up. Is Intel really a better business than it was 5 years ago though?
 
Any theories why Intel screwed up on 10nm and other foundries didn't? Brian talked about "edge stuff", and "thin cleans" in the earnings call. No idea what that means, but maybe some one here knows:
The last part of your question about whether will it be a 10 or 10-plus-plus or 10-plus I think was your question, the yield improvements that we're making are just that, more focused on yield. So think of them as improvements to the various edge stuff, the lithography stuff, thin cleans (33:54) and things like that in order to really drive the multi-patterning and, in some cases, multi-multi-patterning, where you have four, five, six layers of patterning to produce a feature. It's really about that. They aren't necessarily around performance.
Click to expand...
 
lefty said:
Any theories why Intel screwed up on 10nm and other foundries didn't? Brian talked about "edge stuff", and "thin cleans" in the earnings call. No idea what that means, but maybe some one here knows:
Click to expand...

I've read speculation that it's a yield issue pertaining to their use of cobalt for the lower metals.
 
lefty said:
Any theories why Intel screwed up on 10nm and other foundries didn't? Brian talked about "edge stuff", and "thin cleans" in the earnings call. No idea what that means, but maybe some one here knows:
Click to expand...

The part where he mentions up to six layers of patterning for "multi-multi-patterning" to produce a feature is surprising. In fact, you can refer to the presentations on 2017 Intel Technology and Manufacturing Day, where (It was p. 29 in the 14nm report https://newsroom.intel.com/newsroom.../11/2017/03/Ruth-Brain-2017-Manufacturing.pdf) it was mentioned that the cost of SAQP for Intel was the same as everyone else's LELELE, i.e., three masks for patterning.

It is possible "edge stuff" was misheard "etch stuff"(?)
 
Last edited:
Portland said:
Intel still makes the best product. Yes there's a niche market for amd but for most of us intel is the best. I think the problem is that you can only go so far banging your head against the wall and let's be honest that's how most engineering is done. Everyone's struggling not just intel.
Click to expand...


(a) "The best product" is very much a judgement call. If your most important demand is "runs x86 code" then, sure. But if your most important demand is "powers a smart phone" or "powers a watch" or "powers an IoT camera" then ...
The large and growing throughput computing sector is especially contentious, with alternatives like Centriq making a strong showing in integer, and GPUs likewise for FP.

(b) It's likewise not clear that "everyone" is struggling. AMD were able to leap forward substantially from where they were. Apple is already at 1.5x Intel IPC. ARM makes steady gains with its almost annual processor cadence.
It's certainly possible that AMD have hit their limit, or that the Apple A12 and successors are doomed to the pathetic Intel improvements we have seen over the past decade. But these have not happened yet, so it's a strong claim to insist that they are happening.

Likewise on the process side, the three foundries are not struggling the way Intel is.
 
Lets take another moment to remember that Intel bought Altera based on the premise that the first FPGA that comes to market on the latest process will dominate. Xilinx will be shipping 7nm product this year. At this rate Altera will be shipping 10nm product in, I dunno, maybe 2020ish.
 
Yes BUT Portland, what if your architecture that must run 24/7 is based on INTC-only, because you trusted them, with no plug compatible alternative. Then suddenly throughput takes a 30% hit but you still need to supply services to customers. Adapting an alternative takes time but that time has not passed. Perhaps you too would buy extra chips too, for a while, to stay in business. That is until you can design an alternative (which everyone seems to be doing given the job postings at the likes of FB and GOOG, along with the surprise $500M mask-making investment at TSMC). Then, the inability for INTC to move to 10nm becomes just icing on the cake and the decision to move when feasible gets that much easier. Perhaps we are just in a suspended animation mode right now while testing for alternatives takes place; one possible reason for INTC management not being so excited about H2.
 
I'm going against the intel narrative but they are already producing 10 nm at FPGA and low power applications. It could be a demand issue.

We should be focused on quantum computing and it's potential.
 
Why would demand for 10nm be low? Price? Cost is the rumored reason AAPL is backing off of the TSM 7nm+ after all.
 
In the case of Intel, demand could be low because of cost AND the fact that transistor performance for both 10nm and 10+nm is expected to be lower than what they already have with 14++nm.

For TSMC and GloFo as far as I know they do expect some performance increase out of 7nm, although it will not be as much as a new process used to bring.
 
You must log in or register to reply here.
Back
Top