Intel delays 10nm Cannon Lake processor production to late 2019

AppleInsider

Mac users have an even longer wait for new iMacs and MacBook Pro models using 'Cannon Lake' chips, after Intel confirmed the 10-nanometer processors aiming for high-end computers won't arrive any sooner than the 2019 holiday shopping period.

Advising during the processor manufacturer's second-quarter results conference call, executives confirmed chips using the 10-nanometer process are "on track" for vendors to use them in desktops and notebooks in the second half of 2019. CNBC reports interim CEO Bob Swan specified such systems would be on shelves in time for the holiday season.

Cannon Lake has been repeatedly delayed by Intel, with former CEO Brian Krzanich advising in April's results there were issues achieving suitably high yields of 10nm chips. Rather than attempt to perform high volume production and waste large amounts of wafer in the process, the company would instead work to fix the problems before trying mass production.

Intel has managed to ship a small number of 10nm Cannon Lake chips, but not in high enough quantities and with a wide-enough variety for general use by computer vendors. The Core i3-8121U is a 2.2GHz dual-core chip that rises to 3.2GHz under Turbo Boost, with a low 15-watt thermal design point (TDP) making it suitable for notebooks.

The delay on 10nm chips is not only limited to desktop and notebook computers, as it also applies to servers. Xeon processors using the manufacturing process will apparently arrive shortly after the version intended for consumer devices.

For Apple, Cannon Lake showed promise as the family of processors would include support for LPDDR4 and LPDDR4X, types of memory that are low-powered versions of DDR4. This made the chips strong candidates for use in the MacBook Pro, as it would also natively support 32GB of memory.

Until recently, the MacBook Pro lines were limited to a maximum of 16GB of RAM, restricted because of the amount supported by LPDDR3 included in Intel's chips.

For the 2018 MacBook Pro, Apple instead opted to use DDR4 memory, a more power-hungry version that necessitated extra battery capacity, but allowed Apple to move to offering 32GB of memory in the latest model. It is likely that Apple made the move due to the constant delays from Intel over enabling LPDDR4 support in its chips, and decided on an alternate method to increase the memory limit.

In the call, Intel reported record revenue of $16.96 billion for the quarter, beating analyst expectations and up 15 percent year-on-year. Net income for the quarter reached $5 billion, up 78 percent on the same period last year, with an earnings per share of $1.05.

roake

Lots of headlines lately read, “Intel delays...”

Intel needs to get its ship in order.

tmay

Why does that look more like 2020 and what ARM desktop SOC will Apple introduce before that?

k2kw

It’s time for an Aseries based iBook running either macOS or iOS using Marzipan APPS.
maybe Intel would get their act together if they saw they were endangered of losing tons of sales.

I hope atleast Apple has moved over to their own silicon based on the A series chips for their data centers as they are better for heat and power.

linkman

Intel is running up against the fact that wavelengths of light can get only so short (10 nm is pretty stinking short) and physics is going to win at attempts to get much smaller. A transistor can't get smaller than an atom. Progress on ICs is getting slower and will need some sort of breakthrough (like power reduction, using neural-type processing, memristors, light instead of electricity, etc.) before we can see performance improvement rates that we've experienced over the last 40 years.

macxpress

k2kw said:

It’s time for an Aseries based iBook running either macOS or iOS using Marzipan APPS.
maybe Intel would get their act together if they saw they were endangered of losing tons of sales.

I hope atleast Apple has moved over to their own silicon based on the A series chips for their data centers as they are better for heat and power.

Yep...I'd be very surprised if they weren't working on it now. 

wizard69

k2kw said:

It’s time for an Aseries based iBook running either macOS or iOS using Marzipan APPS.
maybe Intel would get their act together if they saw they were endangered of losing tons of sales.

I hope atleast Apple has moved over to their own silicon based on the A series chips for their data centers as they are better for heat and power.

Intel is already loosing out to AMD as in some respects AMD out performs the Intel offerings.  When AMD is behind the performance curve they are not behind by much and for most people it doesnt matter.   Beyond that AMDs APUs have a much better GPU subsystem.    From what im hearing AMD might also beat Intel to a sub 10nm process.  

AMD of course is only one solution to the Intel problem.   Apple could easily introduce an M series ARM based processor for Macs.  The M would be for Mac of course.  They have already demonstrated decent performance in the constrained environment of iPad so with a few more watts i cant see anybody complaining about performance.  A Mac Book could easily handle an 8 or 10 core ARM based chip plus that chip would have all of Apples goodies.  

My greatest fear with respect to Mac is that Apple doesn't give a damn anymore.    I really cant consider the platform again until they demonstrate that the Mac platform means something to them.  The entire desktop lineup is a joke these days and the laptops are not much better.   It is just hard to do business with a company that doesnt respect its own products much less its customers.    I mean really a 4 year old computer with an unchanged list price.  

DuhSesame

Yep, Intel.  What you’d expect.

bonobob

linkman said:

Intel is running up against the fact that wavelengths of light can get only so short (10 nm is pretty stinking short) and physics is going to win at attempts to get much smaller. A transistor can't get smaller than an atom. Progress on ICs is getting slower and will need some sort of breakthrough (like power reduction, using neural-type processing, memristors, light instead of electricity, etc.) before we can see performance improvement rates that we've experienced over the last 40 years.

On the other hand, Apple's current lineup of iPhones is rocking a 10nm SOC, so physics is not the limitation.  Intel is just behind on the manufacturing tech.

Edit: Missing apostrophe.

Soli

I speculated that there could be another Intel delay in the works if Apple felt the need to release this last batch of compromised 15" MBP. The bright side is this gives Apple a lot more power is making waves with their own silicon.

Soli

bonobob said:

linkman said:

Intel is running up against the fact that wavelengths of light can get only so short (10 nm is pretty stinking short) and physics is going to win at attempts to get much smaller. A transistor can't get smaller than an atom. Progress on ICs is getting slower and will need some sort of breakthrough (like power reduction, using neural-type processing, memristors, light instead of electricity, etc.) before we can see performance improvement rates that we've experienced over the last 40 years.

On the other hand, Apple's current lineup of iPhones is rocking a 10nm SOC, so physics is not the limitation.  Intel is just behind on the manufacturing tech.

Edit: Missing apostrophe.

I don't understand how Intel is measuring in a different way from others, but I'm told that Apple's SoC's at 10nm are actually more inline with 14nm.

davidmalcolm

I'm VERY happy with my quad core 13 inch MacBook Pro! Today is a great time to buy if you want a Mac that's going to be the best one available for a long while. This is the machine that I'm hoping takes me until Apple flat out stops supporting Intel in macOS.

macxpress

wizard69 said:

k2kw said:

It’s time for an Aseries based iBook running either macOS or iOS using Marzipan APPS.
maybe Intel would get their act together if they saw they were endangered of losing tons of sales.

I hope atleast Apple has moved over to their own silicon based on the A series chips for their data centers as they are better for heat and power.

Intel is already loosing out to AMD as in some respects AMD out performs the Intel offerings.  When AMD is behind the performance curve they are not behind by much and for most people it doesnt matter.   Beyond that AMDs APUs have a much better GPU subsystem.    From what im hearing AMD might also beat Intel to a sub 10nm process.  

AMD of course is only one solution to the Intel problem.   Apple could easily introduce an M series ARM based processor for Macs.  The M would be for Mac of course.  They have already demonstrated decent performance in the constrained environment of iPad so with a few more watts i cant see anybody complaining about performance.  A Mac Book could easily handle an 8 or 10 core ARM based chip plus that chip would have all of Apples goodies.  

My greatest fear with respect to Mac is that Apple doesn't give a damn anymore.    I really cant consider the platform again until they demonstrate that the Mac platform means something to them.  The entire desktop lineup is a joke these days and the laptops are not much better.   It is just hard to do business with a company that doesnt respect its own products much less its customers.    I mean really a 4 year old computer with an unchanged list price.  

Well to be honest...its not really as great of an investment as it was 15 yrs ago when it was Apple Computer, Inc. That doesn't mean Apple doesn't a give a shit about its Mac platform. If it didn't, it would just shut it down and not develop macOS anymore. I think a lot of the delays are because of Intel, but I do still have a feeling that Apple has something up it's sleeves for the Mac in regards to its own CPU's. It would be a HUGE step for them, both hardware and software and they can double-down on security with things like the T2 chip. 

Any PC manufacturer should be feeling the same way. Dell and HP aren't really getting great returns on their investments in the PC space either. 

Fatman

Although they began the microprocessor revolution, they never were the best. They’ve lost their recipe. And I still blame their inefficient, hot processors that are in nearly every PC and data center for playing a big role in global warming. Hopefully their new radio chip doesnt suck.

wood1208

If Intel able to provide enough 10nm processors early in 2nd half for computer manufacturers to release over holiday season. For, Apple it might be 2020 Spring/summer upgrade cycle with new chassis design,keyboard and internals. Who know, Apple might switch to it's ARM CPU/GPU.

gmgravytrain

linkman said:

Intel is running up against the fact that wavelengths of light can get only so short (10 nm is pretty stinking short) and physics is going to win at attempts to get much smaller. A transistor can't get smaller than an atom. Progress on ICs is getting slower and will need some sort of breakthrough (like power reduction, using neural-type processing, memristors, light instead of electricity, etc.) before we can see performance improvement rates that we've experienced over the last 40 years.

Actually, there are working 5nm transistors that should be going into production in 2020. It's just amazing. https://techcrunch.com/2017/06/05/ibm-creates-a-new-transistor-type-for-5nm-silicon-chips/ I'd heard if we could go to some other materials from silicon, transistor gates could be made much smaller without having an electron tunneling problem.

knowitall

linkman said:

Intel is running up against the fact that wavelengths of light can get only so short (10 nm is pretty stinking short) and physics is going to win at attempts to get much smaller. A transistor can't get smaller than an atom. Progress on ICs is getting slower and will need some sort of breakthrough (like power reduction, using neural-type processing, memristors, light instead of electricity, etc.) before we can see performance improvement rates that we've experienced over the last 40 years.

An atom is a 50th of 10nm, so a lot of room to grow (eh, shrink).
Feature size can be half that of the wavelength, so 5nm should be possible.
Power reduction is a consequence of reduced feature size and no breakthrough.
Neural processing on chip is a special layout of transistors and has nothing to do with speed (and feature size) perse.
Computing with light is related to speed, not feature size.
Memristors have to do with ram ...

gmgravytrain

Soli said:

bonobob said:

linkman said:

Intel is running up against the fact that wavelengths of light can get only so short (10 nm is pretty stinking short) and physics is going to win at attempts to get much smaller. A transistor can't get smaller than an atom. Progress on ICs is getting slower and will need some sort of breakthrough (like power reduction, using neural-type processing, memristors, light instead of electricity, etc.) before we can see performance improvement rates that we've experienced over the last 40 years.

On the other hand, Apple's current lineup of iPhones is rocking a 10nm SOC, so physics is not the limitation.  Intel is just behind on the manufacturing tech.

Edit: Missing apostrophe.

I don't understand how Intel is measuring in a different way from others, but I'm told that Apple's SoC's at 10nm are actually more inline with 14nm.

I'd seen some youtube article telling about how nanometer designs can measured differently, so one of a larger size can be equivalent to another of a smaller size. I think it has to do with the way the gates are designed or just liberal ways of measuring gate size. This video explains it in some detail:

Rayz2016

Yup, thought so. 

https://forums.appleinsider.com/discussion/comment/3077098#Comment_3077098

One thing though: this machine has “stopgap” written all over it. My guess (and it is just a guess) is that Apple was going to continue with 2017 model until Intel delivered the real mobile chip, but they now have reason to believe that Intel will deliver later than they originally thought.

Given how much time it would have taken to redesign the internals and test The 2018 MacBook Pro, I reckon Apple knew about this delay maybe six months ago?

And when this chip does arrive, the gains are going to be nowhere near what they were expecting. 

The switch to home-grown processors can’t come soon enough. 

ascii

Even without 10nm they can make their chips better by improving the architecture or adding more cores. I think Intel will keep pumping out new chips and Apple will be able to keep releasing updated Macs if they want to.

fastasleep

wizard69 said:

My greatest fear with respect to Mac is that Apple doesn't give a damn anymore.    I really cant consider the platform again until they demonstrate that the Mac platform means something to them.  The entire desktop lineup is a joke these days and the laptops are not much better.   It is just hard to do business with a company that doesnt respect its own products much less its customers.    I mean really a 4 year old computer with an unchanged list price.  

JFC, this again. The *only* question mark in the desktop lineup right now is the mini. The iMac is great, and due for a refresh any time now. The iMac Pro is great. The Mac Pro we know is coming next year, and we know they're perhaps even working too hard to get it right this time. "The laptops are not much better" ... We get 32GB of RAM, an improved keyboard, much faster hexacore processors, and people still bitch. Not to mention Mojave shows they care quite a bit about macOS.