Discussion Intel current and future Lakes & Rapids thread

Page 314 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Whats stopping them from shipping high volume 10nm server parts then?

They are struggling with 10nm process sure. It's not like their problem will be solved if they go for chiplets as long as the base process sucks.

Are the new SF features also going to be found in Intel's 7nm nodes? Is it likely the first 7nm won't have the SF features?

New processes always had the feature introduced in previous generations. Strained Silicon from 90nm, Hi-K from 45nm, FinFET from 22nm, its all there in 10nm. They are fundamental to enabling continued scaling. You need the previous steps to go up the next step.
 

positivedoppler

Golden Member
Apr 30, 2012
1,103
171
106
Weren't there rumors that the end of 2021 was the drop dead date for Aurora? If not, perhaps the DOE should consider dropping Intel and goi
I would not say Charlie has been "accurate". More like the title of his rag "semi-accurate". He seem to go by the theory that if you spread enough FUD, some if it will come true.

He's been pretty spot on so far regarding Intel's 10nm delay.
 
  • Like
Reactions: spursindonesia

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
Regarding Charlie's claims, I haven't heard anything about significant recent server delays. Sapphire Rapids might have been pushed out a bit (extra stepping?), but still within the 2021 timeframe, and Granite Rapids is apparently a bit of an organizational mess, but that's nothing new.
 

Hulk

Diamond Member
Oct 9, 1999
4,191
1,975
136
Tiger Lake SuperFin is currently reaching 4.8GHz. Considering that is a mobile part with a low power envelope it seems as though Intel's 10nm SuperFin is doing pretty well, right? Sunny Cove at 4.8GHz would most likely beat Skylake at 5.3GHz in many applications so I can only assume they are going another round at 14+++ because they don't have 10nm SuperFin capacity for mobile and desktop at this point in time.

But if they can move Willow Cove to 10nm Enhanced Superfin at 5+ GHz, those parts could be a reasonable competitor for Zen 3 if there is die space for more cores. Especially if they can reach mid 5GHz clocks on all cores. Intel is definitely down but they ain't dead yet.
 

positivedoppler

Golden Member
Apr 30, 2012
1,103
171
106
I recall he stated over a year ago that the current 10nm on the roadmap was canned and the new 10nm will be nothing like what they originally advertised.

Here are the old articles.

I think people can judge themselves how accurate or inaccurate they were.

BTW, his latest tweet is scary.
regarding when Intel will be competitive in servers again.
"2025 best case"
 
Last edited:

cortexa99

Senior member
Jul 2, 2018
318
505
136
Here are the old articles.

I think people can judge themselves how accurate or inaccurate they were.

BTW, his latest tweet is scary.
regarding when Intel will be competitive in servers again.
"2025 best case"

Charlie might be the earliest one that describe the bad picture of intel's 10nm, and looks like what he said in 2018 were frightening similar to nowaday situation......hard to imagine if intel still cannot deliver on time next year, not only AMD's Zen3/Zen4 but also Apple has its own chips in Mac, time is running out for intel.....
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136

My 3900x scores a 513 in CBR20 ST @ 4.4 GHz (static). Downscaling the Rocket Lake-S ES chip to 4.4 GHz gives an ST score of 542 - a whopping 5.7% faster than my Matisse chip. Hmm.

Tiger Lake SuperFin is currently reaching 4.8GHz. Considering that is a mobile part with a low power envelope it seems as though Intel's 10nm SuperFin is doing pretty well, right?

No idea on yields. Intel still hasn't released anything other than 4c Tiger Lake on this process. Something is amiss.

Sunny Cove at 4.8GHz would most likely beat Skylake at 5.3GHz in many applications so I can only assume they are going another round at 14+++ because they don't have 10nm SuperFin capacity for mobile and desktop at this point in time.

. . . that might be true, but I see no reason why Intel would even think about Sunny Cove on 10SF (or 10SFE).
 
  • Love
Reactions: spursindonesia

Hulk

Diamond Member
Oct 9, 1999
4,191
1,975
136
. . . that might be true, but I see no reason why Intel would even think about Sunny Cove on 10SF (or 10SFE).

The only reason I was thinking Intel might go with Sunny Cove instead of Willow Cove at first crack for 10SF/post Skylake core is the same (likely) reason they are using Sunny vs Willow for Rocket Lake. That being the larger L2/L3 die space is better spent on additional cores.

Of course all of this is being "forced" by Zen3 to a degree. If they really need the extra IPC of Willow Cove (vs. Zen 3) then they may need to sacrifice some profit (die space) and use Willow Cove. Also, could it be that Sunny Cove is more finely tuned at this point to reach higher frequencies than Willow Cove?
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Of course all of this is being "forced" by Zen3 to a degree.

The question is whether Intel can pivot to any new product besides Rocket Lake-S regardless of what is forcing their hand.

Also, could it be that Sunny Cove is more finely tuned at this point to reach higher frequencies than Willow Cove?

Doubtful.
 

RTX

Member
Nov 5, 2020
90
40
61
Did they ever port the SF features for Rocketlake too? The power draw wouldn't be significantly higher, right? Cfv^2 = watts?
 

cortexa99

Senior member
Jul 2, 2018
318
505
136
Ok, only 6 points more for 5.5ghz Rocket Lake or 600mhz higher CPU frequency.IPC is very impresive......................vs Zen 3/4.9ghz:relieved:

If this RKL ST score is true, situation might change a bit but not that much(less core, potentially high power drain and high cost), I hope Xe IGP with 4-6 cores SKUs would have a better value than APUs because it's very likely that mid-range is the only segment RKL can shine.
OTOH seems there's almost no IPC loss when doing backport to 14nm.
 

Hulk

Diamond Member
Oct 9, 1999
4,191
1,975
136
If this RKL ST score is true, situation might change a bit but not that much(less core, potentially high power drain and high cost), I hope Xe IGP with 4-6 cores SKUs would have a better value than APUs because it's very likely that mid-range is the only segment RKL can shine.
OTOH seems there's almost no IPC loss when doing backport to 14nm.

If RKL is somewhat competitive with Zen3 then Intel can make the product "better" by lowering cost.
As Anand used to say, "There are no bad products, only bad price points."
 
  • Like
Reactions: lightmanek

mikk

Diamond Member
May 15, 2012
4,111
2,105
136
The scaling is a bit off on this, at 5.0 Ghz it scored 616 according to Raichu.
 

Asterox

Golden Member
May 15, 2012
1,026
1,775
136
If RKL is somewhat competitive with Zen3 then Intel can make the product "better" by lowering cost.
As Anand used to say, "There are no bad products, only bad price points."

Look how is that going in real life, or Intel 6/12 vs AMD 6/12 CPU king of today Desktop market.

Intel CPU is cheeper, but hm 99% people buy only R5 3600.





R5 3600, 103 000 sold

i5 two versions, 800 sold


"Rocket Lake 8/16 CPU estimated sales", Intel 2 vs AMD 8/16 CPU 20 units no doubt.