[SemiAccurate] Intel kills off the 10nm process

[NTMBK]

SemiAccurate has learned that Intel just pulled the plug on their struggling 10nm process. Before you jump to conclusions, we think this is both the right thing to do and a good thing for the company.

https://semiaccurate.com/2018/10/22/intel-kills-off-the-10nm-process/

Sadly the meat of the story is behind the firewall... I wonder what this means?

1. Aggressive original 10nm is dead, less aggressive "12nm" is new plan
2. All 10nm is dead, Intel skipping straight to 7nm
3. Intel is going to stick with 14nm forever and go fabless for future products (least likely IMO)

 

[coercitiv]

A bit of a shocker, personally I'll wait until Intel official reacts to this claim (silence will also be telling).

 

[Dayman1225]

Man I sure wish I had $1000 laying around to see the details behind that paywall...

My bets are on number two, since one would already assume the original version was killed in favor of the "new version" right? So I am taking this as any and every version has been killed. Though I cannot say for sure without paying $1000

 

[LTC8K6]

The speculation that Intel might skip 10nm and go straight to 7nm has been on this board for a very long time.
It's been discussed a lot here.

 

[jpiniero]

Would it be possible to get a decent shrink using the 14nm tooling?

 

[ttechf]

I think at this point Intel should just skip to 7nm. Even if that means it takes them to the end of 2019 or beginning of 2020 to do it. Perhaps it's just how I feel and I'm not trying to be an "Intel fanboy" or anything but an Intel product at 4Ghz is faster than an AMD product at 4Ghz and always has been. So if Intel is on common ground with AMD [7nm], I honestly think Intel wipes the floor with AMD but it's to be seen and all speculation. If AMD can "turbo clock" to 4.8Ghz or 4.9Ghz on their upcoming 7nm process, then it will make it very difficult for Intel since the last thing I remember reading about Intel and 10nm process was when THEY DO release it you wouldn't see no 5Ghz clockspeeds, most likely around 4.2Ghz to 4.5Ghz. So again, we'll see!

 

[LTC8K6]

I don't think there's any question that if Intel does skip 10nm, they will go to 7nm.

I can't see them going with any other step in-between.

 

[TheGiant]

Option 4- Intel kills the 10nm for its own chips, but has big deal with apple on other stuff
Option 5 - Intel will produce only their chips for small core count (like that famous cannonlake 2C/4T)

Can please someone logically explain what is messed up with Intel 10nm? lots was written but it is too technical for me, something simple or with analogy

How long can take Intel to get to their own 7nm? do we have samples already ?

 

[jpiniero]

Can please someone logically explain what is messed up with Intel 10nm? lots was written but it is too technical for me, something simple or with analogy

Basically they went too dense without EUV. Problem is EUV has problems of it's own and it's unlikely Intel would be ready even in 2020.

The Arizona fab I believe is supposed to be completed construction in 2020, and of course it would take time to get things up and running.

 

[Asterox]

https://semiaccurate.com/2018/10/22/intel-kills-off-the-10nm-process/

Sadly the meat of the story is behind the firewall... I wonder what this means?

1. Aggressive original 10nm is dead, less aggressive "12nm" is new plan
2. All 10nm is dead, Intel skipping straight to 7nm
3. Intel is going to stick with 14nm forever and go fabless for future products (least likely IMO)

It is logical move, green in 2019 and red in future or 2020 and so on.

 

[NTMBK]

Option 4- Intel kills the 10nm for its own chips, but has big deal with apple on other stuff
Option 5 - Intel will produce only their chips for small core count (like that famous cannonlake 2C/4T)

Can please someone logically explain what is messed up with Intel 10nm? lots was written but it is too technical for me, something simple or with analogy

How long can take Intel to get to their own 7nm? do we have samples already ?

I don't think Apple would be dumb enough to risk their product pipeline on Intel's constantly delayed processes- they've been getting progressively worse at hitting schedules since 22nm.

 

[LTC8K6]

Intel won’t reveal anything about its road map, saying through a spokesperson, “We are committed to bringing EUV into production as soon as the technology is ready at an effective cost.” But VLSI Research analyst G. Dan Hutcheson points out that Intel has purchased more EUV tools than any other company.

https://spectrum.ieee.org/semicondu...hography-finally-ready-for-chip-manufacturing

 

[Gideon]

I think at this point Intel should just skip to 7nm. Even if that means it takes them to the end of 2019 or beginning of 2020 to do it.

The thing is, that 10nm was already scheduled for end of 2019, beginning of 2020. If the article is true, it's almost impossible for 7nm to be ready anywhere near that (initial risk production at best). Second half of 2020 is much more likely.

Perhaps it's just how I feel and I'm not trying to be an "Intel fanboy" or anything but an Intel product at 4Ghz is faster than an AMD product at 4Ghz and always has been.

This most certainly was not true during Athlon 64 days, mor Athlon XP actually (though not by that extent)

So if Intel is on common ground with AMD [7nm], I honestly think Intel wipes the floor with AMD but it's to be seen and all speculation. If AMD can "turbo clock" to 4.8Ghz or 4.9Ghz on their upcoming 7nm process, then it will make it very difficult for Intel since the last thing I remember reading about Intel and 10nm process was when THEY DO release it you wouldn't see no 5Ghz clockspeeds, most likely around 4.2Ghz to 4.5Ghz. So again, we'll see!

Process names (7nm, etc) are just marketing by now. Whatever Intel had planned as 7nm (at least before the shakeup) was probably much closer to TSMCs 5nm, so not quite apples-to-apples.

Sorry, if this comes up as nitpicking (isn't intended as such), just couldn't stop pointing those things out

 

[NTMBK]

This most certainly was not true during Athlon 64 days, mor Athlon XP actually (though not by that extent)

Was it even true during the Phenom days? I thought that Phenom had higher IPC than Core 2, just didn't clock as high.

 

[IntelUser2000]

People, people.

If Intel's struggling immensely on the 10nm process, do you think they'll be better off skipping to a 7nm process? Technology improvements are iterative.

 

[epsilon84]

Wow! Fascinating to see how this pans out. How confident are you guys of this being legit?

Was it even true during the Phenom days? I thought that Phenom had higher IPC than Core 2, just didn't clock as high.

Nah, Core 2 had higher IPC: https://www.anandtech.com/show/2378

 

[jpiniero]

Wow! Fascinating to see how this pans out. How confident are you guys of this being legit?

The timing makes sense with Intel's earnings call coming soon. They will probably announce the new CEO too it looks like.

 

[LTC8K6]

People, people.

If Intel's struggling immensely on the 10nm process, do you think they'll be better off skipping to a 7nm process? Technology improvements are iterative.

Intel was probably working on 7nm in the background, so they would not be starting from scratch.

 

[jpiniero]

People, people.

If Intel's struggling immensely on the 10nm process, do you think they'll be better off skipping to a 7nm process? Technology improvements are iterative.

7 is likely very different with EUV involved.

 

[LTC8K6]

Just a couple weeks ago:

Intel May Have 10nm Hardware In-Market Faster Than Expected

https://www.extremetech.com/computi...-10nm-hardware-in-market-faster-than-expected

 

[PotatoWithEarsOnSide]

Is there a comedy section to this forum?
I want to start a thread called "I'm not an Intel fanboy but...(insert your own fanboy comment here)."
In the interests of fairness there'd also be a thread called "I'm not an AMD fanboy but...(insert your own fanboy comment here)."

On topic: any unusual activity recorded in relation to Intel shares leading up to, and then after that SemiAccurate link was first published?

 

[IntelUser2000]

7 is likely very different with EUV involved.

It doesn't matter. EUV makes making dense circuits easier, but you need material innovations to improve performance and lower power use. You might as well expect a first year university student to start working on microprocessors at Intel.

You think in terms of technology, but its really about the people working on the project. If you are struggling on a project, adding more features or making it more advanced won't make it easier for you. You can "skip" it if you are several generations behind and can basically copy others, but this is not the case. Bleeding edge technology is a step by step improvement. Each advance not only gets harder, but brings less gains.

 

[Gideon]

As a user in reddit mentioned, we'll know soon enough: link

Intel earnings date is 10/25

Intel will either state what their 10nm plans are or some analyst will ask them about it

so wait 3 days to get proper information

 

[LTC8K6]

As a user in reddit mentioned, we'll know soon enough: link

Wait?

You've clearly forgotten what forum you're posting on.

 

[dacostafilipe]

If I had a say, I would announce the cancellation of 10nm followed by the announcement of a "new" 9nm for 2020 ...


... right?