Discussion Intel current and future Lakes & Rapids thread

[DrMrLordX]

Q4 2019 for Icelake S.

Ugh let's hope not. That would be bad juju, even for AMD fans.

 

[IntelUser2000]

Ugh let's hope not. That would be bad juju, even for AMD fans.

I am saying how it is. They have problems bringing in a 15W this year so they have to make yet another 14nm part, but people somehow expect a 4.5GHz+ 8 core part to come in late 2018?

There are 3 scenarios for Desktop:
1. Intel has to counter whatever chip AMD has this year, so we'll see an 8 core of some sort. I bet that is Coffeelake based. Is that going to be a "refresh" where it costs the same as 6 core Coffeelake, but costs the same in comes in Q4, which is a year after 6 core Coffeelake?
2. The counter comes in the form of full Coffeelake release with the 300-series chipsets Q2 of 2018. Then they can announce an 8 core is coming at even higher price, in maybe 1-2 months.
3. Do nothing on the desktop space in 2018, until Icelake S is ready. When is that? Q1? Q2? No-one knows. They'll be a sitting duck for 18 months though.

For 1, and 2, Icelake would come 9-12 months later. Which makes it firmly into H2 of 2019.

In the microprocessor space(includes all of them, GPUs/CPUs/FPGA, etc), its always worse than the roadmaps. As some engineers have said, this is really really hard.

I thought the mess Intel had with Pentium III chips were bad. Then they go release Pentium 4's. It can always go worse. Considering what they have to work with, I'd say they are managing it fairly well.

 

[french toast]

I am saying how it is. They have problems bringing in a 15W this year so they have to make yet another 14nm part, but people somehow expect a 4.5GHz+ 8 core part to come in late 2018?

There are 3 scenarios for Desktop:
1. Intel has to counter whatever chip AMD has this year, so we'll see an 8 core of some sort. I bet that is Coffeelake based. Is that going to be a "refresh" where it costs the same as 6 core Coffeelake, but costs the same in comes in Q4, which is a year after 6 core Coffeelake?
2. The counter comes in the form of full Coffeelake release with the 300-series chipsets Q2 of 2018. Then they can announce an 8 core is coming at even higher price, in maybe 1-2 months.
3. Do nothing on the desktop space in 2018, until Icelake S is ready. When is that? Q1? Q2? No-one knows. They'll be a sitting duck for 18 months though.

For 1, and 2, Icelake would come 9-12 months later. Which makes it firmly into H2 of 2019.

In the microprocessor space(includes all of them, GPUs/CPUs/FPGA, etc), its always worse than the roadmaps. As some engineers have said, this is really really hard.

I thought the mess Intel had with Pentium III chips were bad. Then they go release Pentium 4's. It can always go worse. Considering what they have to work with, I'd say they are managing it fairly well.

You have pointed me in the right direction.
If they bring an 8 core coffeelake s? H2 next year that will help them compete, however for that they would need a whole new die, these things take longer than 12 months even for intel, we would have had some leak by now?

If pinnacle ridge delivers any kind of IPC and fabric improvement alongside 10-15% frequency bump intel is in the dog house, with or without coffeelake s 8 core.

 

[Dayman1225]

PCWatch wrote Whiskey Lake is based on 14nm++ (OEM source). From where did you get 14nm+++?



https://pc.watch.impress.co.jp/docs/column/ubiq/1099371.html

Charlie from Semi-Accurate heavily hinted at it and someone else that I trust has said to me the same thing, whether it comes true or not is another thing.

Which meant, wait for it, when that time rolled around we get… 14nm Whiskey Lake, another ‘new architecture’ and ‘new generation of Core’! Be still my beating heart. Rumors about a 14nm part after Whiskey are making the rounds now too. One aside to think about, Intel’s official claims versus reality. Officially these 14nm products are on 14+, 14++, and 14+++ processes which are, again officially, big steps forward.

 

[mikk]

But not for Whiskey Lake. His hint goes towards another 14nm part after WHL which then would make more sense in 14nm+++. But there is no sign for another 14nm part after WHL at the moment, so both might turn out to be wrong.

 

[FIVR]

How is it possible that the new CL prototype it only has 2.2Ghz turbo and 15W TDP? That can't be right, unless there are very serious problems with Intel's 10nm process that go far past yield issues.


They already have Dual cores 2+2 KL-Y chips that do 3.6Ghz boost with a 7W TDP. This would be a massive regression in performance. It would mean that 10nm is performing on par with 22nm in terms of TDP and clock speed. What went wrong?

 

[jpiniero]

How is it possible that the new CL prototype it only has 2.2Ghz turbo and 15W TDP? That can't be right, unless there are very serious problems with Intel's 10nm process that go far past yield issues.

It's not a prototype. It's going to final production. My guess is that it will be an Celeron or Atom branded product for the embedded market. Because of the awful yield Intel is probably binning it as loose as possible and 2.2 Ghz at 15W is what they came up with.

 

[DrMrLordX]

1. Intel has to counter whatever chip AMD has this year, so we'll see an 8 core of some sort. I bet that is Coffeelake based.

Most of the rumours regarding 8c Coffeelake in 2018 seem to have flamed out. Unless Intel is doing a very good job of information/rumour control, I do not fully expect to see 8c Coffeelake.

 

[Qwertilot]

The answer is fairly simple - physics is starting to bite back hard, very hard.

The fundamental laws of the universe aren't trivial wrestling partners.


Sent from my iPhone using Tapatalk

 

[mikk]

Brian Krzanich abandons ship? Sell $54M worth of stock and options

ruh roh

The reason could be this: http://www.businessinsider.de/intel...fter-company-was-informed-of-chip-flaw-2018-1


With this background it makes sense. This is really fishy. Maybe the worst Intel CEO of all times.

 

[Dayman1225]

The reason could be this: http://www.businessinsider.de/intel...fter-company-was-informed-of-chip-flaw-2018-1


With this background it makes sense. This is really fishy. Maybe the worst Intel CEO of all times.

Hope he finally goes....

 

[William Gaatjes]

I do have to write, these are state of the art researchers that have google backing the funds for discovering flaws and potentially using these flaws in cpu's with exploits.
It is not like everybody can devise just an exploit. Although it is serious, i doubt Intel is doing it deliberately. It is a huge company , and the bigger they get, the slower they move.

 

[jpiniero]

BTW, BK's CES keynote is still on for Monday. Should be good.

 

[DrMrLordX]

BTW, BK's CES keynote is still on for Monday. Should be good.

Pity Ajit Pai won't also be there, lulz

 

[witeken]

For those hoping for 10nm soon, don't get your hopes up too high. Intel just released its 10nm yield.

https://twitter.com/witeken/status/950527763299135488

 

[StinkyPinky]

Intel are really struggling here. Time for a new leader i think.

 

[PeterScott]

For those hoping for 10nm soon, don't get your hopes up too high. Intel just released its 10nm yield.

https://twitter.com/witeken/status/950527763299135488

That doesn't look made up at all.

 

[witeken]

That doesn't look made up at all.

Just like it doesn't look like the infamous Itanium chart at all.

 

[elpokor]

Just like it doesn't look like the infamous Itanium chart at all.

lol not at all

 

[Dayman1225]

Intel Mentions 10nm, Briefly

Intel Announces 10nm Is Shipping

Intel claims to have shipped to 10nm chips OEMs in H2'17-2018 and is on track for the 2018 ramp.

 

[Dayman1225]

"I see. So I'm going to assume the new'ish roadmap I saw is correct. No Cannon Lake, just skip over to Ice Lake Y parts in Q4." - David Schor on Twitter

Just spoke with Brian Krzanich. Nothing to add on 10nm: it was shipped to customers by end of 2017. - Ian Cutress on Twitter

 

[Dayman1225]

Intel at CES 2018: 10nm [@8:35] - AnandTech on Youtube.

 

[mikk]

Intel Mentions 10nm, Briefly

Intel Announces 10nm Is Shipping

Intel claims to have shipped to 10nm chips OEMs in H2'17-2018 and is on track for the 2018 ramp.

On track lol. They don't tell us the truth about CNL and 10nm which is embarrassing, only this PR crap.

Just spoke with Brian Krzanich. Nothing to add on 10nm: it was shipped to customers by end of 2017. - Ian Cutress on Twitter

Brian Krzanich is such a clown. But to be honst it was expected to me that he claims CNL is on track.

 

[Dayman1225]

On track lol. They don't tell us the truth about CNL and 10nm which is embarrassing, only this PR crap.




Brian Krzanich is such a clown. But to be honst it was expected to me that he claims CNL is on track.

Well if they actually did "ship" to "customers" they can claim whatever they like, like being on track according to the H2'17 Ship and 2018 ramp timeline.

 

[LTC8K6]

If they are vulnerable to meltdown, then I don't think I want to buy them.