Intel's 10nm Is Broken, Delayed Until 2019

[PaulAlcorn]

Intel announced its financial results today, and although it posted yet another record quarter, Intel unveiled serious production problems with its 10nm process.

Intel's 10nm Is Broken, Delayed Until 2019 : Read more

 

[InvalidError]

Yet another 14nm refinement year, yay!

Intel used to be a solid two years ahead of everyone else, that lead shrunk to a year on 16/14nm, now it looks like Intel might be falling a year behind on 10/7nm.

This will hopefully have interesting consequences on competitiveness and pricing pressure.

 

[dudmont]

70% performance improvement in the 14nm node? How much of that is entirely from adding cores?
Intel used to be worth 8% IPC improvement to each new series. It looks to me like AMD is about 1 generation away from true parity and they still have em beat in cores and threads by a good amount. Is all the flurry in hiring a sign of someone hitting the panic button?
Good luck in Houston . Here's hoping my Vikes get their G.

 

[modeonoff]

Does that mean CPUs such as Cascade Lake that should have the Meltdown and Spectre issues removed at hardware level will be delay to 2019?

 

[rhysiam]

So let me get this straight: competition from AMD and issues with the 10nm process meant that Intel, fuelled by a caffeine hit, rushed "Coffee Lake" to market. This was to remain competitive until they could come out all-guns-blazing with the true successor, "Cannon Lake". Now instead, as 10nm problems persist, we'll all have to drown our sorrows with "Whisky Lake".

Is that about the measure of it?

 

[InvalidError]

Is that about the measure of it?

Hopefully the ice lake will be strong enough not to break when Intel gets on it.

 

[electro_neanderthal]

Meh, no big deal. 5nm is basically the smallest architecture achievable in a normal environment according to the laws of physics (or something like that), and both Intel and AMD are going to struggle the closer they get to that point. AMD just hasn't hit that wall yet, but it's coming.

I imagine one day we'll have processors in vacuum sealed glass tubes with small AC systems inside to go any smaller. Bring on the "Edison Core" (it'll look so steampunk)!

 

[jpe1701]

Why wouldn't Intel announce that they have z390and an 8 core cpu coming at this. Unless they don't. It just seems logical because it would hurt Ryzen sales by people that would wait, and reassure investors. Why the secrecy?

 

[PaulAlcorn]

Why wouldn't Intel announce that they have z390and an 8 core cpu coming at this. Unless they don't. It just seems logical because it would hurt Ryzen sales by people that would wait, and reassure investors. Why the secrecy?

Intel brought out Coffee Lake and had really low volume during the holiday season, so you basically couldn't buy one. And when you could, they were insanely overpriced. That likely hurt Intel's sales in that period because no one would buy a Kaby with fewer cores than the pending Coffee Lake models. Retailers ran deep discount Ryzen sales during Black Friday and Cyber Monday, so AMD owned the holiday shopping season.

Announcing an eight-core could have a similar impact on Intel's Coffee Lake sales. Why buy a six-core when an eight-core is right around the corner? Meanwhile, AMD has really competitive silicon in the market. Pre-announcing products is a tricky business. Sometimes it hurts more than it helps.

 

[canadianvice]

Amdahl's law is a bitch.

 

[SkyBill40]

They just snatched Jim Keller from AMD, so that should certainly help Intel along or so I would assume. At this rate, Intel will be lucky to hit 10nm by 2020.

 

[Arbie]

"It just seems logical because it would hurt Ryzen sales by people that would wait"

True theoretically - but nobody with any sense would wait. Ryzen now thrashes the comparable Intel parts in every important way. No, I'm not talking about 95 vs 100 FPS in some game. Plus AMD has saved us from the wasteland Intel promised for the future. That alone should earn them our business. They are certainly getting mine.

 

[Arbie]

BTW, excellent article, thanks.

 

[Nintendork]

AMD Zen2 will surpass Intel cpu's IPC by a good margin and will be unchallenged in 2019.

 

[Nintendork]

Cannon Lake is not even the solution to fight Zen, it was only a die shrink from Skylake. The true competitor was supposed to be Ice Lake, their new arch after Skylake which is nowhere to be seen.

 

[rantoc]

I just love that the important CEO keynotes are missing, point 2 was that they admittedly funneled a lot of money from the very prosperous PC division towards the failing ones, its nothing new there but sadly for them they can't milk as much because they just got competition and are forced to reinvest at least some money in the area and start to innovate again.

What do we have to thank for this? AMD ofc, they came around swinging and have a really solid cpu/platform, bought an 2700x + x470 (asus crosshair vii hero) and it's frankly been stabler and felt less half-baked than the 7700k + z270 (asus strix) that i replaced in my rig.

AMD does really deserve a prais here or we would have had 20 more years of no innovation but ofc at higher prices.

 

[InvalidError]

AMD Zen2 will surpass Intel cpu's IPC by a good margin and will be unchallenged in 2019.

Improving per-core IPC is easier said than done, that's why the focus for all CPU designers is on more cores as software capable of using them becomes more common instead of higher IPC and clock frequency at any cost. There is a limit to how much instruction-level parallelism (ILP) can be extracted from typical software's instruction flow and the cost of approaching that limit increases quite rapidly.

At best, Zen 2 / Ryzen 3 might go from slightly worse per-core IPC than Intel on average to slightly better. Any pressure on Intel is ultimately a win for everyone else.

 

[noveed]

Should have called the next gen 14nm processors Acid Lake, sounds more fun.

 

[LordVile]

70% performance improvement in the 14nm node? How much of that is entirely from adding cores?
Intel used to be worth 8% IPC improvement to each new series. It looks to me like AMD is about 1 generation away from true parity and they still have em beat in cores and threads by a good amount. Is all the flurry in hiring a sign of someone hitting the panic button?
Good luck in Houston . Here's hoping my Vikes get their G.

AMD are more power efficient as well. With everything up to the 1700 having a 65W TDP and intel 8600K having 95W

Wait... What did I just say?

 

[redgarl]

They just snatched Jim Keller from AMD, so that should certainly help Intel along or so I would assume. At this rate, Intel will be lucky to hit 10nm by 2020.

HAHA, you are funny, Jim Keller left before the release of Zen... typical fanboy glasses...

 

[The Empress]

Production problem = not cheap enough to offset cannibalizing ourselves.

 

[feelinfroggy777]

MERGED QUESTION
Question from modeonoff : "Does Intel's 10nm is broken means delay of Cascade Lake to 2019?"

Hello, as Intel plans to remove meltdown and Spectre issues at hardware level of CascadeLake CPUs, I was making some plans based on that. With this announcement, does that mean CPUs with these vulnerabilities removed at hardware level won’t be out until next year and it is not certain which month that will be?

It is rumored that these vulnerabilities will be removed with their next release. So a delay would mean it would slow that release. With that being said, there are only a very few specific situations and applications where the OS update and microcode update to address the vulnerabilities where it will slow your CPU performance significantly. Most process performance differences are within the margin of error and most users will never even notice.

 

[KidHorn]

Things just keep looking up for AMD. In this business that doesn't last long.

 

[modeonoff]

So given that the availability of such bug removed CPUs won’t be out until next year (could be earlier, mid of Fall 2019), if I need a temporary system to do CUDA computations using GPU, it is better to build a Threadripper than a 8700K system now?

 

[SkyBill40]

They just snatched Jim Keller from AMD, so that should certainly help Intel along or so I would assume. At this rate, Intel will be lucky to hit 10nm by 2020.

HAHA, you are funny, Jim Keller left before the release of Zen... typical fanboy glasses...

How utterly astute of you, smart guy.

From the story talking about Keller's hire by Intel on this very website:

"More importantly, Keller was the lead architect of AMD's Zen microarchitecture while he was the company's corporate vice president and chief cores architect."

I'm well aware he wasn't at AMD any longer when Intel hired him; however, he helped with the creation of Zen prior to its release and then moved on. So what difference does his not being employed by AMD at the time Intel got him make? Oh, right: NONE. He did work at AMD for quite a while and that work is probably what he is best known for, so your statement doesn't amount to much other than being a typically worthless ad hominem move on your part. Well done. Try focusing on facts and context next time rather than resorting to the juvenile... unless that's what suits you best.