Discussion Intel current and future Lakes & Rapids thread

Page 214 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Richie Rich

Senior member
Jul 28, 2019
470
229
76
Alder Lake is pretty likely not coming out until H1 22.
For 2022 is planned Golden Cove at 7nm. It seems to me like products collision. It's more like a wish of AMD fan than a reasonable possibility. Much more probable is Alder Lake in 2021. Golden Cove uarch was originally planned for 2021 at 7nm so it must be ready to deploy at 10nm too.

Rocket Lake and Alder Lake share one common thing: they are using older process node than planned. IMHO these could be first visible Jim Keller's moves since he's in Intel for two years already. Preparing CPU at different node can take about two years. Also big+little concept of Alder Lake is close to his ARM background from Apple. Intel caught AMD unprepared once (Core2Duo vs. K8). It can happen again, especially with Keller on board ;)
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,686
136
They REALLY shouldn't even bother with RKL
Intel got a lot of flak for their lack of contingency plans for the 10nm fiasco. We criticized them for being too aggressive with the node, for not increasing core count sooner, for not back-porting sooner.

RKL is the safe approach. Safe equals some degree of redundacy. If they have something better at the time, it can be used as filler and faded away from the spotlight. If not, would you rather see Skylake Eternal Edition?
 
  • Like
Reactions: lightmanek

Richie Rich

Senior member
Jul 28, 2019
470
229
76
No, 2022's product (Meteor Lake) should be Ocean Cove. There's some disagreement out there about what is exactly Ocean Cove, but it's Ocean Cove one way or another.
I thought Ocean Cove with Austin team was cancelled. Maybe you mean Ocean Cove code name as Golden Cove successor. Ok, that makes sense. So this year 14nm RKL, 2021 10nm Golden Cove in Alder Lake and 7nm Ocean Cove in Meteor lake in 2022.
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
Intel got a lot of flak for their lack of contingency plans for the 10nm fiasco. We criticized them for being too aggressive with the node, for not increasing core count sooner, for not back-porting sooner.

RKL is the safe approach. Safe equals some degree of redundacy. If they have something better at the time, it can be used as filler and faded away from the spotlight. If not, would you rather see Skylake Eternal Edition?
No, I'd rather see a roadmap that is not intentionally full of straight-on lies. Because I'm sure that the real roadmap that we never get to see, makes some sort of sense.
Nothing we can officially hear does though.
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
Ocean Cove is dead as both a name and a project. Intel will hopefully host another architecture day where they reveal the next two names after Golden Cove and Gracemont.

As for Meteor Lake, I suspect the fabs will be the least of their problems.
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,686
136
No, I'd rather see a roadmap that is not intentionally full of straight-on lies. Because I'm sure that the real roadmap that we never get to see, makes some sort of sense.
Nothing we can officially hear does though.
Huh? It's one thing to doubt their roadmap in terms of feasibility, another thing entirely to substitute theirs with yours.
 

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
Intel got a lot of flak for their lack of contingency plans for the 10nm fiasco. We criticized them for being too aggressive with the node, for not increasing core count sooner, for not back-porting sooner.

RKL is the safe approach. Safe equals some degree of redundacy. If they have something better at the time, it can be used as filler and faded away from the spotlight. If not, would you rather see Skylake Eternal Edition?
Intel may well have contracts to provide RKL, under whatever SKU, to OEMs, especially given the slide it showed up on.

-------------------------------------------------

As far as long term for Intel, if we don't start hearing something about 7nm in H2, I'd be getting pretty nervous.
 

uzzi38

Platinum Member
Oct 16, 2019
2,565
5,575
146
Ocean Cove is dead as both a name and a project. Intel will hopefully host another architecture day where they reveal the next two names after Golden Cove and Gracemont.

As for Meteor Lake, I suspect the fabs will be the least of their problems.

I didn't think it was dead in name as well. Hmm.

Well in any case, agreed on the second half of that. I highly doubt 7nm will cause any issues for Intel.
 

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
I didn't think it was dead in name as well. Hmm.

Well in any case, agreed on the second half of that. I highly doubt 7nm will cause any issues for Intel.

What's annoying me is that we have no freaking idea how Intel 7nm is doing - not a clue. So, I remain of the opinion that we can't say 7nm won't cause Intel problems.
One can always hope that the rumors of 10nm desktop chips is a bait and switch tactic, but hope in this case is not a virtue.
 
  • Like
Reactions: scannall

OriAr

Member
Feb 1, 2019
63
35
91
The fact that we haven't heard of any issues yet have me thinking that at the very least it's going OK, no 10nm like problems at least.
And PVC is apparently sampling late this year, that should mean something....
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
Huh? It's one thing to doubt their roadmap in terms of feasibility, another thing entirely to substitute theirs with yours.
I'm not sure if you're being serious here, but if yes, I'll just leave it at that, because my English knowledge is obviously too lackluster to express myself in a way that you understand too. If not, good laughs man, good laughs :)
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,686
136
I'm not sure if you're being serious here, but if yes, I'll just leave it at that, because my English knowledge is obviously too lackluster to express myself in a way that you understand too. If not, good laughs man, good laughs :)
What I meant was there is a big difference between inferring someone is lying and inferring the truth. A lie detector is not a truth detector, hence my advice to refrain from reconstructing the puzzle with so many missing pieces.

You seem very tempted to guess Intel real plans, I advised you against it. Other than that, cheers and all, this isn't my native language either, so we might be lost in translations anyway. :p
 
  • Like
Reactions: spursindonesia

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
What I meant was there is a big difference between inferring someone is lying and inferring the truth. A lie detector is not a truth detector, hence my advice to refrain from reconstructing the puzzle with so many missing pieces.

You seem very tempted to guess Intel real plans, I advised you against it. Other than that, cheers and all, this isn't my native language either, so we might be lost in translations anyway. :p
I can't help it if that's your take. cheers
 
  • Like
Reactions: spursindonesia

amrnuke

Golden Member
Apr 24, 2019
1,181
1,772
136
The fact that we haven't heard of any issues yet have me thinking that at the very least it's going OK, no 10nm like problems at least.
And PVC is apparently sampling late this year, that should mean something....
Very good news indeed. I really hope they ramp things up quick.

What's annoying me is that we have no freaking idea how Intel 7nm is doing - not a clue. So, I remain of the opinion that we can't say 7nm won't cause Intel problems.
One can always hope that the rumors of 10nm desktop chips is a bait and switch tactic, but hope in this case is not a virtue.
I think that's tough. Even N5 at TSMC hasn't entered big time mass production with the major domestic vendors in the US (at least, nothing that's been announced publicly). So Intel's 7nm density being on par with TSMC's 5nm, and given how far behind they've been with process shrinks in their fabs, I think either they've shifted a lot of resources to 7nm, at the expense of 10nm, and 7nm may be ready sooner than expected, and the quiet on the news front is secrecy / focus on 10nm --- or the reason there's no news is because they're still focusing major efforts on 10nm and there just isn't much news on 7nm period.
 

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
Very good news indeed. I really hope they ramp things up quick.


I think that's tough. Even N5 at TSMC hasn't entered big time mass production with the major domestic vendors in the US (at least, nothing that's been announced publicly). So Intel's 7nm density being on par with TSMC's 5nm, and given how far behind they've been with process shrinks in their fabs, I think either they've shifted a lot of resources to 7nm, at the expense of 10nm, and 7nm may be ready sooner than expected, and the quiet on the news front is secrecy / focus on 10nm --- or the reason there's no news is because they're still focusing major efforts on 10nm and there just isn't much news on 7nm period.
Uh, thought Apple's A14 was due on N5, which is presently ramping up. Am I wrong?
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
Uh, thought Apple's A14 was due on N5, which is presently ramping up. Am I wrong?
You aren't wrong:
"We also believe that meaningful wafer starts for new iPhones should commence in April - May timeframe and TSMC is unlikely to see a meaningful bottleneck in production process," writes Hariharan. "We believe that the 5nm process node has been ramping up reasonably smoothly through 4Q19 and 1Q20." ==> https://appleinsider.com/articles/20/03/25/tsmc-a14-chip-wont-be-holdup-for-5g-iphone-12
"TSMC is set to kick off volume production of chips built using 5nm process technology in April, and has already seen the process capacity fully booked by clients, according to industry sources." ==> https://www.digitimes.com/news/a20200311PD200.html

For amusement only, TSMC hasn't really updated their website:
=> It is scheduled to start risk production in the second half of 2019.
=> Our 5nm technology entered risk production in March 2019...

Money does really solve node issues.
 
Last edited:

ondma

Platinum Member
Mar 18, 2018
2,718
1,278
136
You aren't wrong:
"We also believe that meaningful wafer starts for new iPhones should commence in April - May timeframe and TSMC is unlikely to see a meaningful bottleneck in production process," writes Hariharan. "We believe that the 5nm process node has been ramping up reasonably smoothly through 4Q19 and 1Q20." ==> https://appleinsider.com/articles/20/03/25/tsmc-a14-chip-wont-be-holdup-for-5g-iphone-12
"TSMC is set to kick off volume production of chips built using 5nm process technology in April, and has already seen the process capacity fully booked by clients, according to industry sources." ==> https://www.digitimes.com/news/a20200311PD200.html

For amusement only, TSMC hasn't really updated their website:
=> It is scheduled to start risk production in the second half of 2019.
=> Our 5nm technology entered risk production in March 2019...

Money does really solve node issues.
Except for Intel 10 nm. I still cant believe a company with such expertise and resources could screw up so badly, and after what, 3 years or so, has not really solved the issues.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,480
14,434
136
Except for Intel 10 nm. I still cant believe a company with such expertise and resources could screw up so badly, and after what, 3 years or so, has not really solved the issues.
You have probably never worked for a big company (>100,000 employees) You would be appalled at what goes on. So many people should have been fired over this, and most likely never were. They are still trying to work things out.
 
Mar 11, 2004
23,031
5,495
146
The fact that we haven't heard of any issues yet have me thinking that at the very least it's going OK, no 10nm like problems at least.
And PVC is apparently sampling late this year, that should mean something....

I wouldn't read that into it at all. Which, I doubt its going as bad as 10nm apparently was/is, but I wouldn't read silence as things are peachy. Time will tell and we'll just have to wait and see.

I will say that their fab here in Chandler (which I believe is their lead 7nm fab) still had construction cranes up, but I'm not sure anything much can be read into that.
 

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
I wouldn't read that into it at all. Which, I doubt its going as bad as 10nm apparently was/is, but I wouldn't read silence as things are peachy. Time will tell and we'll just have to wait and see.

I will say that their fab here in Chandler (which I believe is their lead 7nm fab) still had construction cranes up, but I'm not sure anything much can be read into that.
Hmm, I wonder. I thought Intel built a new fab in Arizona just for 7nm (or held it empty for 7nm). Really wish we had more 'inside' info.

Edit: Gosh darn it, was I drunk yesterday - obviously Chandler is in Arizona. o_O
 
Last edited:

uzzi38

Platinum Member
Oct 16, 2019
2,565
5,575
146
Okay, things got a little weird last night.

First of all - living proof that Intel were planning on taking CNL-H and CNL-S up to 8 cores (there was a LinkedIn page snippet before, but this is actual proof):


And then second, is something that related to a funny story that floated around a while back. Now I'm sure everyone knows that Gen10 graphics were broken. Completely and utterly broken. Well, the story said that Intel actually took the time to manually fix a broken chip. Like, by hand. I know.

Well, the idea doesn't seem so farfetched now: