14nm 6th Time Over: Intel Readies 10-core "Comet Lake" Die to Preempt "Zen 2" AM4

[mikk]

Yeah, other posters already talked about the hypothetical "double ring" design before Comet Lake-S 10C was mentioned by "chrisdar".

Yes exactly, the double ringbus posting didn't come from chrisdar which means we can ignore it. It doesn't even make sense because we know Intel doesn't need a double ring design for 10C. The extra complexity, chip size and power consumption isn't worth the effort for a mainstream chip with just two additional cores.

 

[IntelUser2000]

I didn't think it could happen but Intel keeps proving me wrong. Now there's a chance that Icelake won't make it in volume in 2019 except in extremely limited quantities like Broadwell in 2014.

Imagine how dominant Intel's process group was before this fall. No single person with enough power to influence decisions ever thought 10nm will be delayed, despite 14nm being delayed for 6-9 months or so.

So they said: "Let's save money and NOT have a 14nm backup plan!. What a brilliant idea! We'll gain 2% more margins this year by not having a backup plan. SURELY our process teams can NEVER fail."

This is why its never about the tech, but always about the people. Sure, good tech is necessary, but who develops them? Smart PEOPLE!

 

[Spartak]

Looks like i'll need to eat some crow if all intel has for the desktop in 2019 will be another skylake 14nm iteration. Three years of failure and still no contingency plan by backporting icelake to 14nm if 10nm isnt ready yet?

Seriously i dont think i've ever seen this level of incompetence before.

Well, Its reminiscent of AMD being unable to iterate on athlon 64 and totally squander their lead.

 

[IntelUser2000]

Brian Krzanich happened.

Just another reminder how important a CEO is.

There was an article about how Paul Otellini was going to fire Kraznich but Kraznich turned the tables around and got Otellini to resign instead.

The same author was talking about how Kraznich's team was going for false diversity. That is, pretend you are all about diversity but really you are going the opposite. Kraznich fired many women in the management team, some of which were said to be real good at their jobs, but he probably didn't want anyone against him in any way.

Really, anyone else might have been better. Kraznich happened because the previously expected successor to Otellini, Sean Maloney, had serious health problems. The decision became between Kraznich and David Perlmutter. It was said the BoD thought they wanted an American and Perlmutter wasn't so Kraznich got it instead. Then Perlmutter got sacked once Kraznich was CEO.

But then again, Intel didn't have an excellent CEO since Craig Barrett. That's like 25 years of problems compounding on each other. At least Barrett and Otellini were sane.

 

[scannall]

There was an article about how Paul Otellini was going to fire Kraznich but Kraznich turned the tables around and got Otellini to resign instead.

The same author was talking about how Kraznich's team was going for false diversity. That is, pretend you are all about diversity but really you are going the opposite. Kraznich fired many women in the management team, some of which were said to be real good at their jobs, but he probably didn't want anyone against him in any way.

Really, anyone else might have been better. Kraznich happened because the previously expected successor to Otellini, Sean Maloney, had serious health problems. The decision became between Kraznich and David Perlmutter. It was said the BoD thought they wanted an American and Perlmutter wasn't so Kraznich got it instead. Then Perlmutter got sacked once Kraznich was CEO.

But then again, Intel didn't have an excellent CEO since Craig Barrett. That's like 25 years of problems compounding on each other. At least Barrett and Otellini were sane.

One of Otellini's HUGE mistakes was turning down Apple for making their iPhone SOC. Forcing Apple to do their own made Apple into a giant brain drain on the CPU design engineers (Have a look at their linkedin sometime), plus a massive infusion of cash into the pay for play foundries. In any management at Intel still has a lick of sense they can steal some CPU talent from Global Foundries since they faceplanted with IBM's golden gift. But they better be quick, they aren't the only one who wants/needs talent.

 

[jpiniero]

Looks like i'll need to eat some crow if all intel has for the desktop in 2019 will be another skylake 14nm iteration. Three years of failure and still no contingency plan by backporting icelake to 14nm if 10nm isnt ready yet?

For desktop, that's been the rumor for some time. It's possible that the Icelake mobile being released is on a custom 14 nmish process, although it would obviously would have to be derived from the SoC variant or something that would give a decent density/power consumption improvement. But we'll have to see.

 

[The Stilt]

We (the consumers) don't need any more cores, at least for the time being...

 

[Markfw]

We (the consumers) don't need any more cores, at least for the time being...

Speak for yourself. US DC'ers need all the cores we can get.

 

[LTC8K6]

Speak for yourself. US DC'ers need all the cores we can get.

But you probably don't need 10 cores at the sort of power and heat that 14nm is going to require. Also the cost, given the 9900K numbers.

 

[Markfw]

But you probably don't need 10 cores at the sort of power and heat that 14nm is going to require. Also the cost, given the 9900K numbers.

Well, my new 16 core threadripper is what, 12nm ? and does not produce as much heat as the 9900k, and for $410 ?

 

[scannall]

Well, my new 16 core threadripper is what, 12nm ? and does not produce as much heat as the 9900k, and for $410 ?

For your DC purposes the Threadripper is a much better choice, for a lot of reasons.

 

[EXCellR8]

$410?

 

[LTC8K6]

Well, my new 16 core threadripper is what, 12nm ? and does not produce as much heat as the 9900k, and for $410 ?

Sure, unless Intel pulls a rabbit out of it's hat, I can't see how this supposed 10 core 14nm chip is going to make any sense.
It's got to be hotter and more power hungry than the 9900K.
It seems very unlikely to work in the current motherboards.
Maybe they can leave out the IGP and save some wattage?

The only way it makes sense is at 10nm.

The 9900K at least made some sense, trying to hold on to the desktop gaming crown.

We have the 9820X and the 9900X at 14nm if we want to spend more money and deal with higher TDP for 10 cores with Intel.

 

[Markfw]

$410?

It was that last week on ebay, and $450 on newegg. Tuesday or wed.

 

[StinkyPinky]

You'd think Intel would have at least some understanding of what AMD will be offering with Zen 2. If this 10 core cpu really is 14nm it suggests to me Intel know AMD are going to release a 10+ core mainstream CPU and they are desperate to match it. Sure it will melt and throttle like hell, but Intel can still market it as a 10 core cpu

 

[EXCellR8]

It was that last week on ebay, and $450 on newegg. Tuesday or wed.

not bad

 

[Markfw]

You'd think Intel would have at least some understanding of what AMD will be offering with Zen 2. If this 10 core cpu really is 14nm it suggests to me Intel know AMD are going to release a 10+ core mainstream CPU and they are desperate to match it. Sure it will melt and throttle like hell, but Intel can still market it as a 10 core cpu

And do you think it will be less than AMD's threadripper with 16 cores ? I doubt it. 8700k is good for gaming, 9700k is even better. But a 10 core will be a waste for mainstream. If you need that many cores, threadripper is a better value.

 

[moonbogg]

I think 10 cores on mainstream are about two too many. 6/12 is currently perfect and 8/16 is a new standard with plenty of headroom for growth. Oh, but then again Intel doesn't make 6/12 or 8/16 anymore, so never mind. I don't consider a $500 8/16 chip to be a reasonable choice for mainstream these days, so I don't count the 9900K. It's a stupid chip because of a stupid price.

Oh! Maybe they will give us 10/10 as perhaps a reflection of how they think they are doing these days when faced with actual competition. I say Intel gives us a 10/10 and we should give them a 10/10 for their effort. A for effort Intel. Watch them do a 10/10 chip, haha.

 

[Burpo]

$500? $399 with discount code EMCXEERV2

https://www.newegg.com/Product/Prod...ProcessorsDesktops-_-19117958-S0U&ignorebbr=1

 

[crashtech]

$500? $399 with discount code EMCXEERV2

https://www.newegg.com/Product/Prod...ProcessorsDesktops-_-19117958-S0U&ignorebbr=1

If you can find me a 9900K for $400, I will likely buy it. That link takes me to an 8C/8T 9700K.

 

[Markfw]

$500? $399 with discount code EMCXEERV2

https://www.newegg.com/Product/Prod...ProcessorsDesktops-_-19117958-S0U&ignorebbr=1

Why not a 16 core 32 thread for $449 ? Instead of 8 core 8 thread.? If you are a gamer and want to pay the bucksss. A 8700k would be cheaper for a gamer, and just fine.
https://www.newegg.com/Product/Prod...tion=1950x&cm_re=1950x-_-19-113-447-_-Product

Edit.... wait thats $405 right now.... At Amazon, $370

 

[jpiniero]

If it is branded 10th gen, then yes Comet Lake will likely be a new socket. But it probably would be an orphan given that Tigerlake would likely use the McIVR, even on desktop. Unless of course the 2020 desktop is also Skylake based...

Do think it's still possible that they would drop the IGP to GT1 for the 10 core die, if only to reduce die space to maintain margins. I'm assuming the price is intended to be similar to the 9900K.

 

[Roland00Address]

What the hell is happening to Intel? Yes I know what is happening, but the reason for the previous sentences is to make a point on how astonished I am about how bad it is getting. Intel is already supply constrained on 14nm, yet they must make larger cpu dies for AMD is competing in their high margin brackets so Intel has to make the larger cpu dies to compete. This is in turn cannibalizing sells for they are already supply constrained.

All the while AMD is causing them to have downward pricing with various skus. This is a vicious cycle.

Now this was foreseeable with how long 10nm has been delayed, but at the same time knowing something intellectually is about to happen and seeing the next 6 months or 12 month future. Well this is astonishing, it is getting this bad.

 

[PhonakV30]

so 10 cores like 2x 5 cores in CCX module , like Zen coes ?

 

[coercitiv]

so 10 cores like 2x 5 cores in CCX module , like Zen coes ?

The source of the leak made no mention of a 2x5 core configuration, some news outlet(s) conflated the info from the leaker with someone else's comment on a possible dual bus design and went on a strange path from there.

So, to sum it all up, if the leak is genuine all we know for now is Comet Lake is a 10C part and Intel revises their roadmaps every 3 months.