bit-tech.net

Intel confirms 10nm Cannonlake delay, tick-tock miss

Intel confirms 10nm Cannonlake delay, tick-tock miss

Intel chief Brian Krzanich has confirmed rumours that the company's 10nm Cannonlake parts are being pushed out to 2017, with the 14nm Kaby Lake to take its place.

Intel has announced that it has suffered a hiccough in its tick-tock development cycle, confirming rumours that its upcoming 10nm parts have been delayed and will be replaced by 14nm alternatives.

Having lead the charge to turn Moore's Law, the observation by company co-founder Gordon Moore that the number of transistors in an integrated circuit tends to double every 18 months, into a obey-or-die rule for the semiconductor industry, Intel has based its development on a tick-tock cycle: each 'tick' is a process shrink, reducing the size of components and lithography to continue cramming ever-more transistors into its chips; each 'tock' is a change to the processor's microarchitecture, while leaving the components the same size.

Cracks in the march of Moore's Law appeared when Intel was forced to delay its 14nm parts, admitting that the continued shrinkage was causing yield issues. This was followed by recent rumours that its 10nm node, the next 'tick' in its release cycle, had hit a major roadbump and was being pushed back to at least 2017 - a rumour Intel has now confirmed.

According to details released during its latest earnings call, the 10nm Cannonlake parts - the successor to the upcoming 14nm Skylake - have been pushed back until the second half of 2017, and will be replaced in the interim with additional 14nm parts dubbed Kaby Lake. As part of the announcement, Intel chief Brian Krzanich warned that 'our [tick-tock] cadence today is closer to two-and-a-half years than two,' its previously-roadmapped release schedule.

During the same call, Intel announced lower-than-expected financial drops for the second quarter of the year: revenue was down five per cent to $13.2 billion, operating income down a whopping 25 per cent to $2.9 billion, and its net income down three per cent to $2.7 billion on a 62.5 per cent margin - two percentage points lower than the same period last year. Intel blamed the poor performance of the traditional PC market, and by extension his company's lack of presence in the smartphone and tablet markets, with a 14 per cent drop in revenue for the Client Computing Group. Growth areas included the Data Centre group, which saw revenue rise 10 per cent, and the burgeoning Internet of Things division, which increased its revenue four per cent to a still-small $559 million.

Krzanich also confirmed that the first Skylake parts are on-schedule for production in the second half of this year, but did not confirm a firm launch date.

9 Comments

Discuss in the forums Reply
Corky42 16th July 2015, 11:36 Quote
Maybe now Intel can invest some of their massive R&D budget into solving other problems and not just chasing ever shrinking node sizes, I'm not saying they have ignored historical problems or haven't been innovative, it's just sometimes being so focused on achieving a specific goal can lead to other things being knocked down the priority list.
rollo 16th July 2015, 11:50 Quote
Like what?

If they cared enough for mobile they would of done something about it a long time before the big 2 players in mobile developed the chips themselves.

Most company's would still kill for Intels Revenue and Income figures for a quarter.

My guess is that Intel do not want to throw massive amounts of cash at 10nm to get the problems with yields resolved in a quick time. And are happy to continue at there current pace of development with little to no Competitors.
Corky42 16th July 2015, 12:29 Quote
Well that's an unanswerable question if ever I saw one, I don't work for Intel and even if I did I'm not sure I would be privy to how they've prioritised their R&D budget.

I was merely suggesting that maybe by trying to stick to Moore's Law that they may have neglected some other projects, research, developments, or problems, by trying to solve the inevitable problem that came with every node shrink they have perhaps neglected other areas of CPU design.
schmidtbag 16th July 2015, 13:41 Quote
Quote:
Originally Posted by Corky42
Well that's an unanswerable question if ever I saw one, I don't work for Intel and even if I did I'm not sure I would be privy to how they've prioritised their R&D budget.

I was merely suggesting that maybe by trying to stick to Moore's Law that they may have neglected some other projects, research, developments, or problems, by trying to solve the inevitable problem that came with every node shrink they have perhaps neglected other areas of CPU design.

If this were AMD we were talking about, or maybe a company like TSMC, then I think you'd be onto something. But intel has more money than they know what to do with. They actually are hiring more engineers and sponsor 3rd party projects.

But generally, I agree that Intel is hardly innovative and today they don't have to do anything to get customers. They haven't had a compelling product since Sandy Bridge and yet they're still making huge amounts of revenue.
rollo 16th July 2015, 15:28 Quote
Well you did suggest they have other problems. I was simply wondering what they were.
Corky42 16th July 2015, 16:21 Quote
Quote:
Originally Posted by rollo
Well you did suggest they have other problems. I was simply wondering what they were.

Maybe a bad choice of words, sorry.

Would focusing on over-looked performance improvements possibly ignored in the past because of the need to get the node shrink done instead, have been a better word to use? :p
silk186 16th July 2015, 17:22 Quote
I'm not concerned, always more they can improve: native usb 3.1 support, more cores, more lanes...
megamale 17th July 2015, 10:51 Quote
Quote:
Originally Posted by silk186
I'm not concerned, always more they can improve: native usb 3.1 support, more cores, more lanes...

This.

Since the i7 920 I have never upgraded for lack of CPU performance. It has always been the platform like lanes, sata3, native usb3, etc... I certainly couldn't care less about further power savings on the desktop...
Gareth Halfacree 17th July 2015, 10:55 Quote
For those wanting to see Intel really innovate, be careful what you wish for: last time Intel tried something really novel, we got NetBurst and the godawful Pentium 4...
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums