Intel 14 nm delayed.

[aigomorla]

Intel's process is their pride and joy and the envy of the industry, always has been. They would never, ever blame it for a delay unless it was and I'm 100% sure that every single Intel fanboy here would agree with that statement.

Original C2D Pentium Line... - Delayed 3 months to tweak yield process.
C2D 6X50 Series.. - Delayed to fix Trusted Computing
Q9000 Yorkfield Series.. was originally delayed due to Bad Stepping
(then intel gave u guys the ever so infamous E0 Stepping.)

Bloomfield - Bad launch... had to do stepping revision 3 months after intial launch which pissed everyone off cuz new stepping dominated overclocking world.
(intel promised not to do this again.... and so far its the only promise they have still kept...)
(intel said they wouldnt make us change boards... however that is BS... )

Gulftown - Delayed 3 weeks due to failure in A0 stepping in terms of reliability + durability.

I remember the last 2, cuz my NDA's were extended... and i remember the forum raging on gulftown cuz i couldnt post any results due to the NDA extension.
and it pissed everyone off i had to Black out everything... like this..

However every delay from intel has sort of been worth it for our end.
Do you really want another bloomfield repeat? i for sure dont...

So what are u talking about no delays for intel?
Intel is also full of delays.... if intel kept on there original roadmap... we should have Haswell-E.

 

[ShintaiDK]

You are breaking his FUD illusion that Intel is in trouble and dying.

 

[aigomorla]

You are breaking his FUD illusion that Intel is in trouble and dying.

ROFL... shintai do u remember my gulftown launch.. and the thread being full of rage cuz i was under heavy NDA.
Man... where IDC when i need him...

But intel has had its fair share of delays...

 

[SiliconWars]

How many of those did Intel mention in quarterly CC's?

 

[aigomorla]

How many of those did Intel mention in quarterly CC's?

they mentioned them to the entire tech world.

First off when a CPU is launched, stores like newegg are not the first places to get them.

The OEM manufactorers, the big guys get the first pick at cpu's in batch 1000.
When a product line is delayed, they announce it offically to let all those OEM members know there is a problem with yield of production, and a delay will be involved.
99% of the time.. the big guys get sample ES's... they also get NDA extension notices.... and all other sorts... the delay is not hidden...

Infact in the last 2 generations, intel has been nothing but delays...
Why do u think my faith / trust in them has crumbled to almost nothing in the past 3 years?

Not to mention, they are not as flexable as they used to be when we overclockers didnt exist. I personally liked it better when we didnt exist to intel.

Because i can tell you, we wouldnt have a enthusiast platform which makes gives them an excuse to neuter the consumer platform.

 

[SiliconWars]

Yes but my point is, they have done it here. They've finally admitted the issue to the investors, what the rumours had long been saying.

I'm well aware of Intel's slips, they are always happening but generally never go longer than 3 months. That is why this one is different - the fact that they've not been able to keep the illusion of everything being perfect going when they did in the past.

They *had* to release this information at the CC because they ran out of time to hide the bad news and get it fixed for certain. Broadwell is very late - much later than intended, and that's ~6 months minimum.

 

[Revolution 11]

They are certainly not in trouble but they are sure slipping off the two year tick-tock timeframe with each node.

First 45nm product release on Q4 2007 (October 28 2007) Core 2 Extreme QX9650 (Penryn)
First 32nm product release on Q1 2010 (January 4 2010) Core i5 and Core i3 (Clarkdale) ------> ~27 months
First 22nm Product release on Q2 2012 (April 23 2012) Core i5 and i7 (IvyBridge) ------> ~28 months
First 14nm product release on Q4 2014 (October 2014 ???) Broadwel (first time Mobile product gets the new process before Desktop) ------> ~30 months

According to this, they have never been strictly on a 2 year cadence. More like 2 years and 4 months. And this will be 3 months more. I expect them to stay on the 2.5 year cadence until 7 nm probably.

I also expect there to be many delays, both with Intel and anyone else who is left to pursue Intel all the way to 7 nm.

When Intel says in 2012 that 5 nm could be out by 2019, I automatically add 2 years to this estimate. Because that's probably closer to the actual release date.

 

[aigomorla]

Yes but my point is, they have done it here. They've finally admitted the issue to the investors, what the rumours had long been saying.

thats because companies like apple brought snobby investors who needed to be hand fed all the reports instead of digging them out themselves.

 

[AtenRa]

According to this, they have never been strictly on a 2 year cadence. More like 2 years and 4 months. And this will be 3 months more. I expect them to stay on the 2.5 year cadence until 7 nm probably.

I also expect there to be many delays, both with Intel and anyone else who is left to pursue Intel all the way to 7 nm.

When Intel says in 2012 that 5 nm could be out by 2019, I automatically add 2 years to this estimate. Because that's probably closer to the actual release date.

Im expecting everyone (Intel included) to be at a 3 year cadence very soon. It seams AMD and NVIDIA will be the first with GPUs at 28nm. The lower the volume of the products, the faster they will turn into the 3 year cadence.

 

[SiliconWars]

Im expecting everyone (Intel included) to be at a 3 year cadence very soon. It seams AMD and NVIDIA will be the first with GPUs at 28nm. The lower the volume of the products, the faster they will turn into the 3 year cadence.

I mostly agree. Intel might stay a couple of months faster but everybody including them is slowing down.

I can't find the link now but I do believe that Intel said they weren't slowing down a year, year and a half ago? There's no magic involved though and the laws of physics are what they are.

 

[Abwx]

It's fixed. Feel free to try and play word games with what has been publicly stated though, since apparently it wasn't adequately declarative. Doesn't change the fact that the primary issue has been resolved. My guess is that they waited until now to say anything for the simple fact that announcing that there's been a problem without stating that it's been resolved is far more detrimental than staying quiet in the interim.

Why do I have confidence? Because, we have got back now and added additional fixes, gotten back onto that curve, so we have confidence that the problem is fixed, because we have actually data and defects and so that gives us the confidence that we are to keep moving forward now and that happens sometimes in these development phases like this, so that's why we are going to over it a quarter.

http://techreport.com/news/25518/intel-pushes-broadwell-production-into-early-2014

Three time the word "confidence" in just a few lines ,
at the end this sound like an incantation...

 

[SiliconWars]

How anyone can read that and proclaim it's fixed for sure is beyond me. He could barely make it clearer that this is still in the lap of the gods.

 

[Fjodor2001]

Im expecting everyone (Intel included) to be at a 3 year cadence very soon. It seams AMD and NVIDIA will be the first with GPUs at 28nm. The lower the volume of the products, the faster they will turn into the 3 year cadence.

Yeah, and even if they could make it quicker, why bother? Node shrinks don't bring the frequency increases and much lower power consumption that they used to. The main remaining advantage nowadays is that you can fit more transistors per die area, leading to lower production cost. But that should be weighed against the additional R&D cost needed to stay on the bleeding edge of process technology. To sum it up, the benefit of Intel's process technology advantage is shrinking.

 

[Khato]

How anyone can read that and proclaim it's fixed for sure is beyond me. He could barely make it clearer that this is still in the lap of the gods.

Could there be another hurdle further down the line? Sure, that's the case while tuning any new process node. But for now the defect density curve for 14nm is improving like it should again instead of being stuck.

 

[Exophase]

How anyone can read that and proclaim it's fixed for sure is beyond me. He could barely make it clearer that this is still in the lap of the gods.

Saying that they got the data back and it looks good and he's confident it's fixed now is the strongest thing that can be said. That means they did the fix, they have every reason to believe it's effective, and they're moving forward. You're making it sound like they're still working it out and taking a continual day by day slip. The impact to schedule is over. There could be more impact to schedule later but there's no point speculating about this since it isn't based on anything.

This sort of thing is normal procedure. You find a bug, you apply a fix, you do some analysis and/or minimal testing to verify your fix, and you move forward. There's still a non-zero chance that more extensive testing later could reveal issues, but for right now you've done your due diligence towards fixing it and you express levels of confidence to your boss/client/investors/whatever.

 

[dahorns]

Yes but my point is, they have done it here. They've finally admitted the issue to the investors, what the rumours had long been saying.

I'm well aware of Intel's slips, they are always happening but generally never go longer than 3 months. That is why this one is different - the fact that they've not been able to keep the illusion of everything being perfect going when they did in the past.

They *had* to release this information at the CC because they ran out of time to hide the bad news and get it fixed for certain. Broadwell is very late - much later than intended, and that's ~6 months minimum.

How do you go from Broadwell being delayed a quarter to a ~6 month minimum? You understand what a quarter is right? Technically, we don't even know for sure that the delay in production will affect product delivery timelines. If the vendors believe production issues will be sorted out, there is no particular reason the delay should affect product deployment.

 

[SiliconWars]

A quarter can be "absorbed" without needing to come out with an official statement. Anything after that just can't be ignored.

Look at the link I gave - http://www.semiwiki.com/forum/content/2640-intel-14nm-delayed.html

What I was told is that the Intel 14nm process has not left the OR development facility to be replicated exactly in the OR and AZ fabs.

14nm move-in was supposed to happen in Q3 but it did not. I got an update this week and was told it would “probably” not happen until Q1.

Q3 to "probably" Q1 is 6 months guys. They kept it under wraps as long as they could - ie as long as it was "just" a quarter delay.

Yep, 3 months ago they knew it was "probably" not a near fix. Let me just reiterate though, in bold again, for the 3rd time.

It probably is fixed, I'm just not willing to trust in it.. If I was forced to give a percentage I'd say 70% chance of it being fixed.

You should be prepared for some bad news down the line though. Intel does not idly throw around process issues in investor calls. Ever.

 

[dahorns]

A quarter can be "absorbed" without needing to come out with an official statement. Anything after that just can't be ignored.

Look at the link I gave - http://www.semiwiki.com/forum/content/2640-intel-14nm-delayed.html

Q3 to "probably" Q1 is 6 months guys. They kept it under wraps as long as they could - ie as long as it was "just" a quarter delay.

So you're suggesting that not only did the CEO lie when he said things were on pace for Q4, but that is lying yet again when he says that the delay is only a quarter (i.e. the real delay is 6 months)?

 

[SiliconWars]

So you're suggesting that not only did the CEO lie when he said things were on pace for Q4, but that is lying yet again when he says that the delay is only a quarter (i.e. the real delay is 6 months)?

Q4 is from 1st October to December 31st. December 31st is the endpoint.

A "quarter delay" is making sure it's out before the 31st of March.

That's 6 months from October to March explained in a "quarter" delay.

Please listen - if you can, you will absorb minor issues on yield or whatever, and you can still make your original statement of "quarter 3" or "quarter 1". That's why they give quarterly guidance instead of monthly or weekly.

Intel ran out of time for their "quarter 4". They made that every single time before now, but this time they didn't make it. They came out and admitted it in the CC because they ran out of time. They admitted the problem and like any other company said they believed it was solved.

 

[SOFTengCOMPelec]

Q4 is from 1st October to December 31st. December 31st is the endpoint.

A "quarter delay" is making sure it's out before the 31st of March.

That's 6 months from October to March explained in a "quarter" delay.

1 + 1 = 3

Sorry, I don't get it.

A quarter is 3 months NOT EQUAL to 6 months.

EDIT: I DO get it now, but you are EXAGGERATING the issue, as far as I can tell.

 

[Exophase]

You say 6 months minimum, then justify it with a Q3 to Q1 slip which is 6 month maximum. Talk about seeing things in the worst way possible.

 

[SiliconWars]

You say 6 months minimum, then justify it with a Q3 to Q1 slip which is 6 month maximum. Talk about seeing things in the worst way possible.

You're a really smart guy Exophase but like most of your ilk you just don't understand marketing. And SoFTengCOMPelec as well.

 

[SiliconWars]

Sorry, mistake.

 

[jdubs03]

This whole issue of a quarter delay is nonsense. If anything Intel choreographed the impending delay by saying Broadwell would launch H2 2014. Its a 3-month delay not a 6-month, so we'll be seeing devices coming on shelves around September. This might not be that bad of a thing, maybe more designs will come out at a similar time.

 

[AtenRa]

This whole issue of a quarter delay is nonsense. If anything Intel choreographed the impending delay by saying Broadwell would launch H2 2014. Its a 3-month delay not a 6-month, so we'll be seeing devices coming on shelves around September. This might not be that bad of a thing, maybe more designs will come out at a similar time.

The roadmap was for H2 2014, that is from July 2014 to December 2014. Q3 is from July to September. If they will miss one quarter then the release will be in Q4 IF the original release would be on Q3. That is the release could be from October to December 2014 at the earliest.
Now if they were planing to release in Q4, with the new data we are going for a Q1 2015 release.

So, even if they will release in early Q4 2014(October) retail products will not be ready until Q1 2015.