Intel's 10 nm delayed until 2019

[Johan45]

Appears as if yields are proving difficult.

The closer we get to Cannon Lake, Intel's next major processor update that die shrinks production to 10 nanometers, the further it seems to slip away. Unfortunately, that is the case once again—Intel announced today that it won't be shipping 10nm processors in high volume this year, and instead is now targeting 2019.

My big takeaway on this was:

The significance there is that Intel is able to manufacture 10nm processors, but has more work to do before it can ship Cannon Lake and other 10nm CPUs in volume. In the meantime, Intel plans to continue optimizing its 14nm product lines.

Read more: https://www.pcgamer.com/intels-10nm-cannon-lake-cpus-wont-arrive-until-2019/

Intel announced its financial results today, and although it posted yet another record quarter, the company unveiled serious production problems with its 10nm process. As a result, Intel announced that it is shipping yet more 14nm iterations this year. They'll come as Whiskey Lake processors destined for the desktop and Cascade Lake Xeons for the data center.

Read more: https://www.tomshardware.com/news/intel-cpu-10nm-earnings-amd,36967.html

 

[mackerel]

I look forward to 14nm+++++++++

 

[trents]

I think Intel may be hitting up against the nano meter die shrink wall that has theoretically been predicted to happen for a long time. Time to start looking looking at entirely different approaches to building processors and doing micro-processing. People have been looking into alternative technologies for a long time but the incentive and the research money has not been there for a serious attempt at it. Maybe that time has come. Maybe silicon is on the way out.

 

[Johan45]

I think Intel may be hitting up against the nano meter die shrink wall that has theoretically been predicted to happen for a long time. Time to start looking looking at entirely different approaches to building processors and doing micro-processing. People have been looking into alternative technologies for a long time but the incentive and the research money has not been there for a serious attempt at it. Maybe that time has come. Maybe silicon is on the way out.

They are EMIB
Not everyone is having problems either. 7nm and even 5 are possible ATM

 

[Alaric]

Gee, wonder why they hired Jim Keller.......

 

[wingman99]

From what I have been reading the other manufactures don't have a standard of a true 5nm and 7nm node measurement.

Krzanich also admitted that the company's density lead over competing fabs is shrinking.

 

[Alaric]

From what I have been reading the other manufactures don't have a standard of a true 5nm and 7nm node measurement.

It seems Team Blue is a long ways off from worrying about that.

 

[Dr. McCoy]

I'm not sure exactly what this means but headway is being made in some form..... But the form in which it happens will determine what's offered once done, if it gets done at all.
https://www.techpowerup.com/forums/...lerator-data-pushing-silicon-deployed.243698/

Who's to say whether this means Intel with the troubles they seem to have with it won't fall behind for the first time in who knows how long - Could be.
But they now have Keller on the job so that's probrably not gonna happen.

 

[Johan45]

AMD also plans to sample 7nm Epyc CPUs this year https://wccftech.com/amd-zen-2-design-complete-7nm-epyc-2018/

During the recent Q4 2017 earnings call, AMD confirmed that the design for their next-generation Zen 2 cores is complete and ready for sampling in late 2018. This would mean that we can see the next-generation Zen cores deployed within products in 2019.

https://www.tomshardware.com/news/amd-earnings-call-tsmc-7nm-gpu,36957.html

AMD CEO Lisa Su revealed that the company has working 7nm GPU silicon in its labs, which we know from its pre-CES announcements is a Radeon Instinct MI25 card with the Vega architecture. AMD says the 7nm GPUs and 7nm EPYC 2 processors are on schedule to sample to customers by the end of the year. Both 7nm products will ship in volume in early 2019.

 

[Alaric]

I'm not sure exactly what this means but headway is being made in some form..... But the form in which it happens will determine what's offered once done, if it gets done at all.
https://www.techpowerup.com/forums/...lerator-data-pushing-silicon-deployed.243698/

Who's to say whether this means Intel with the troubles they seem to have with it won't fall behind for the first time in who knows how long - Could be.
But they now have Keller on the job so that's probrably not gonna happen.

Although, I have to wonder if Keller can/will fix the Intel problems or just say "You can't get there from here" and start from scratch on a new chip. That could give Team Red a year or two of high speed, low drag market share acquisition. No one believes the Big Blue Monolith is going away, but Intel surely did get hit with a perfect storm of lousy timing. Ryzen, followed by Intel's headless chicken routine throwing random chips at the wall trying to get one to stick, leaving their customers screwed on chipset/socket needs, then Spectre/Meltdown and its performance hit, followed by the Ryzen + price/performance ratio. All while the 10 nm process seems to be in Greek for them.

 

[wingman99]

GlobalFoundries does not actually make 12nm node.

The reason the company can do that, almost without question, is 12nm isn’t an actual process node by the old definition at all. It’s the kind of iterative improvement that foundries used to roll out as a half-node, or referred to as a new type of capability on the same process node. https://www.extremetech.com/computi...ew-globalfoundries-12nm-node-future-cpus-gpus

 

[Dr. McCoy]

Although, I have to wonder if Keller can/will fix the Intel problems or just say "You can't get there from here" and start from scratch on a new chip. That could give Team Red a year or two of high speed, low drag market share acquisition. No one believes the Big Blue Monolith is going away, but Intel surely did get hit with a perfect storm of lousy timing. Ryzen, followed by Intel's headless chicken routine throwing random chips at the wall trying to get one to stick, leaving their customers screwed on chipset/socket needs, then Spectre/Meltdown and its performance hit, followed by the Ryzen + price/performance ratio. All while the 10 nm process seems to be in Greek for them.

He may well say just that to them.
I know years ago back when 64 bit was first introduced Intel coudn't make it work - They tried to but everything they did on their own blew up in their face so they finally HAD to make a deal with AMD to learn how it worked and to start making them. AMD was off and running with 64 bit capable chips but Intel was still cranking out 32 bit chips until that happened. Once they did get it they used their deep pockets to run away with the performance crown - I'm sure AMD won't be so easy to deal with this time if another situation like it happens.

The sting of being behind for so long is still fresh in their minds but if enough $$ thrown their way that can make them forget.

 

[Alaric]

He may well say just that to them.
I know years ago back when 64 bit was first introduced Intel coudn't make it work - They tried to but everything they did on their own blew up in their face so they finally HAD to make a deal with AMD to learn how it worked and to start making them. AMD was off and running with 64 bit capable chips but Intel was still cranking out 32 bit chips until that happened. Once they did get it they used their deep pockets to run away with the performance crown - I'm sure AMD won't be so easy to deal with this time if another situation like it happens.

The sting of being behind for so long is still fresh in their minds but if enough $$ thrown their way that can make them forget.

The Itanic! To borrow a phrase I heard from Linus earler. LOL. The Itanium disaster. Intel has made a career of trying to invent proprietary tech to cut everyone else out of the market. They remind me of Bose-"Better sound through marketing". If Intel had spent more resources on an alternative to the rapidly diminishing returns of clock speed increases and wasted less effort on Thunderbolt and Optane they might be in better shape right now.

 

[mackerel]

I think AMD64 allowed short term gains for users, but we might be paying a long term price for it. A native 64-bit CPU build ground up to do it should perform much better than the 64-bit on top of 32-bit on top of 16-bit on top of 8-bit x86 architecture. Just compatibility sucks and here we are. I suppose I shouldn't be surprised, history is littered with lots of examples of the inferior but cheaper solution winning over the better by more expensive solution. The masses want good enough at minimal price. Ryzen is a direct play on that strategy.

 

[Alaric]

I don't know about "better". The Itanium was another Intel shot at at creating a proprietary market they could own. Compatibility is kind of a big deal in the computing eco system. The most amazing, world changing breakthrough in the history of the Universe is a useless paperweight if nobody can use it. It reminds me of the guy who claims the world would be perfect if everyone would just do what he wants. And I think you're selling Ryzen short. It's many times more CPU than the vast majority needs. That's hardly "good enough". Other than a few user specific scenarios, Intel offers nothing that will show an advantage to 90% of the market.

260 million PCs were sold in 2016. The percentage of buyers who could discern any difference between CPUs is pretty small. "Better" is subjective, and for the bulk of the market, Intel's version of "better" is a distinction without a difference.

 

[mackerel]

And I think you're selling Ryzen short. It's many times more CPU than the vast majority needs. That's hardly "good enough". Other than a few user specific scenarios, Intel offers nothing that will show an advantage to 90% of the market.

That is my point. Ryzen doesn't need to match everything Intel does, and by cutting parts it helps lower costs and apparent power rating. Mass buyers go on value not performance. If Intel joins in, we could be on a downward trend.

 

[bmaxa]

I think AMD64 allowed short term gains for users, but we might be paying a long term price for it. A native 64-bit CPU build ground up to do it should perform much better than the 64-bit on top of 32-bit on top of 16-bit on top of 8-bit x86 architecture. Just compatibility sucks and here we are. I suppose I shouldn't be surprised, history is littered with lots of examples of the inferior but cheaper solution winning over the better by more expensive solution. The masses want good enough at minimal price. Ryzen is a direct play on that strategy.

amd64 is native 64 bit CPU. It's not that you don't have alternatives, x86 is simply fastest....

 

[mackerel]

amd64 is native 64 bit CPU. It's not that you don't have alternatives, x86 is simply fastest....

Processors would happily run pure 32-bit too, so in that sense not what I'd consider pure 64-bit.

 

[Alaric]

Value is a performance parameter, too. Intel's efforts to lock others out of the eco system are shady and bad for the market and the consumers. It's an attempt to stifle innovation and eliminate competition. That doesn't foster better performance, it ensures "You'll take what we give you and like it. And if it isn't actually better we'll sabotage the compiler and lie to you about it". Die hard Intel fans should be loving AMD right now. Intel has to put up or shut up, and do so at a competitive price. All hail King Intel.

Long Live AMD.

 

[mackerel]

Value is a performance parameter, too. Intel's efforts to lock others out of the eco system are shady and bad for the market and the consumers. It's an attempt to stifle innovation and eliminate competition. That doesn't foster better performance, it ensures "You'll take what we give you and like it. And if it isn't actually better we'll sabotage the compiler and lie to you about it". Die hard Intel fans should be loving AMD right now. Intel has to put up or shut up, and do so at a competitive price. All hail King Intel.

Long Live AMD.

It could go either way. My concern above is that if they attempt to follow AMD's cost reduction, you WILL get less in future. There has been past talk about AVX-512 going into consumer CPUs, and if they do that, great. But at the same time, they're also talking about increasing performance per watt more so than we've seen, and to me these two goals don't align. They're not exclusive, but it wont be easy to avoid the ongoing negligence by many attempting to compare red and blue CPUs due to their differences.

I still find it amusing that people complain that Intel is too expensive compared to AMD. If AMD does what you want cheaper, get it and why do you care what Intel does? There's a not insignificant number of people who want Intel CPUs at AMD pricing. Actually, so would I, but I'm not expecting it.