• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

[Ashraf] 10nm "Lakefield" SoC with Intel big + little cores

Page 19 - 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
7,092
1,622
136
One thing I don't really understand. Why going big little style? Let's say for hypothetical 1 big 1 little core compared to a dual core of the same architecture CPU with 1 core with transistors optimized for let's say 1.6 GHz max frequency and 1 core with transistors that can run up to 4 GHz? Is the power savings gained from in order architecture that much compared to out of order architecture?
Tremont is not in-order. They moved to out of order in 2013. It's a fairly wide architecture with performance similar to Ivy Bridge.

The way you are suggesting would result in crap multi-threaded performance. Lakefield has 4 Tremont cores for that.

Also the quad Tremont core cluster takes up roughly the same size as a single Sunny Cove core.

How much I/O is integrated into Lakefield that is not in Renoir, but rather in a separate southbridge chip?
I don't know about I/O, they might be similar since Renoir has integrated PCH functions. But Lakefield also has PoP memory which has to be part of TDP and will take up at least 1W.
 

vigilant007

Junior Member
Dec 7, 2014
21
10
81
Tremont is not in-order. They moved to out of order in 2013. It's a fairly wide architecture with performance similar to Ivy Bridge.

The way you are suggesting would result in crap multi-threaded performance. Lakefield has 4 Tremont cores for that.

Also the quad Tremont core cluster takes up roughly the same size as a single Sunny Cove core.



I don't know about I/O, they might be similar since Renoir has integrated PCH functions. But Lakefield also has PoP memory which has to be part of TDP and will take up at least 1W.
Tremont is out of order. It’s also coming from the family that performs the worst at all things for a very long time. Intel actually publicly abandoned the Atom line years ago. It isn’t an exaggeration at all to say that i have a dual core 64 bit processor on my watch that out performs any dual core configuration of Atom today.

I get having 4 is better than 2, and 4 is better than 1 that intel sold in Netbooks. But come on, is anyone really expecting 4 or these cores with a single core of Intels Core line to be better than say the Quad Core variant in MacBook Airs?

We are dealing with Intel literally using verbiage for how Toyota marketed a slow car from 20 years ago to explain the existence of a chip all of us knows will just be garbage.

The Surface Neo will release next year in whatever form, with this ill conceived “hybrid CPU” because “reasons”. It will go up against the entry level iPad, and iPad Air. By then the iPad Air will have an A14. The entry level iPad will be running the A13 or A12. All of us know the Surface Neo is going to probably cost more than the Surface Duo. I’d be shocked if it didn’t. Explain the performance in 2021. Trying to defend the performance today is pretty bad.

Lake field would have made a TON of sense 3 years ago. But it’s released today. As someone that just spent $3k on a new MacBook Pro 16, who is seeing a chip from 2 years ago get within 10% of the performance that I get today in a development board in the DTK has questions.

All of the posturing is “cute” at best. I’m saying this as someone who lives on the Apple ecosystem. I actually really like the Surface Duo and Surface Neo. Looking at what they chose for hardware, I’m fine seeing this bad science experiment play out in public.
 
  • Like
Reactions: dr1337 and Lodix

Thala

Golden Member
Nov 12, 2014
1,051
387
136
Tremont is out of order. It’s also coming from the family that performs the worst at all things for a very long time. Intel actually publicly abandoned the Atom line years ago. It isn’t an exaggeration at all to say that i have a dual core 64 bit processor on my watch that out performs any dual core configuration of Atom today.

I get having 4 is better than 2, and 4 is better than 1 that intel sold in Netbooks. But come on, is anyone really expecting 4 or these cores with a single core of Intels Core line to be better than say the Quad Core variant in MacBook Airs?

We are dealing with Intel literally using verbiage for how Toyota marketed a slow car from 20 years ago to explain the existence of a chip all of us knows will just be garbage.
Yes this thing is garbage when comparing to ARM cores, even when comparing to 2 year old standard ARM Cortex A76 - never mind Apple cores. Thing is some people value compatibility with a large range of Windows apps - which you cannot get easily with ARM based designs.
On the bright side however when you are less reliant on compatibility with all Windows x86/x64 apps at least you have the option today getting an ARM device either with Windows like the Surface Pro X, Galaxy Book S etc. or going with the soon to be released Apple Silicon Macs - few years back and you did not have a choice at all.
 
Last edited:
  • Like
Reactions: vigilant007

vigilant007

Junior Member
Dec 7, 2014
21
10
81
Yes this thing is garbage when comparing to ARM cores, even when comparing to 2 year old standard ARM Cortex A76 - never mind Apple cores. Thing is some people value compatibility with a large range of Windows apps - which you cannot get easily with ARM based designs.
On the bright side however when you are less reliant on compatibility with all Windows x86/x64 apps at least you have the option today getting an ARM device either with Windows like the Surface Pro X, Galaxy Book S etc. or going with the soon to be released Apple Silicon Macs - few years back and you did not have a choice at all.
You’re right.

Many are reliant on those systems. I am reluctantly saying that that is a Microsoft problem.

I can give a lot of props to Microsoft for badly “softly” trying to move things forward.

I’ve been in a very sad room where a Microsoft employee told people that thought there was a path from Silverlight to WinRT (not really a thing) that they were out of luck.

Thats the problem.

Microsoft has literally no control of their stack. Yet others are doing better to move their platform forward.

I know what Microsoft could be. But seriously, they have no control of their stack.

Prove me wrong.
 
Last edited:

dmens

Golden Member
Mar 18, 2005
1,966
285
126
One thing I don't really understand. Why going big little style? Let's say for hypothetical 1 big 1 little core compared to a dual core of the same architecture CPU with 1 core with transistors optimized for let's say 1.6 GHz max frequency and 1 core with transistors that can run up to 4 GHz? Is the power savings gained from in order architecture that much compared to out of order architecture?
Jumping on a tech trend bandwagon without understanding anything. Fairly typical Intel move. Lakefield 1+4 should really be considered a 0+4 because the big core is so utterly crippled from thermal and power limitations.
 

vigilant007

Junior Member
Dec 7, 2014
21
10
81
Full of Hyperbole there.
“Ooooo...Kay...” prove me wrong. Atom processors haven’t done well at anything so I’m more than happy to hear what you have to say.

Please, show me a scenario where Atom smokes anything.

Until then, I check your hyperbole, and I raise you.

I’d bet you’ve got nothing but a hand full of 2, 5, and 7s and you are hoping for some match that make the “full of hyperbole there” to put up at the river.

Atom was dead for years for a reason. Sprinkling glitter and glue on it, and calling it hybrid doesn’t make that dead turkey any more of an eagle.

Intel should have left that dog in the ground, when they really should have cremated it.

It’s always cute when someone says “hyperbole” and not use a reference. Just like Intel’s “Lakefield” architecture, it’s cute.
 
Last edited:

ondma

Golden Member
Mar 18, 2018
1,469
406
106
“Ooooo...Kay...” prove me wrong. Atom processors haven’t done well at anything so I’m more than happy to hear what you have to say.

Please, show me a scenario where Atom smokes anything.

Until then, I check your hyperbole, and I raise you.

I’d bet you’ve got nothing but a hand full of 2, 5, and 7s and you are hoping for some match that make the “full of hyperbole there” to put up at the river.

Atom was dead for years for a reason. Sprinkling glitter and glue on it, and calling it hybrid doesn’t make that dead turkey any more of an eagle.

Intel should have left that dog in the ground, when they really should have cremated it.

It’s always cute when someone says “hyperbole” and not use a reference. Just like Intel’s “Lakefield” architecture, it’s cute.
You are the one making the claim, up to you to provide the proof, instead of just outrageous smack talk. BTW, you also just proved the accusation of hyperbole correct.
 

vigilant007

Junior Member
Dec 7, 2014
21
10
81
You are the one making the claim, up to you to provide the proof, instead of just outrageous smack talk. BTW, you also just proved the accusation of hyperbole correct.
You can say what you want, I have no desire to talk “outrageous smack”.

Intel failed in the market place with Atom. That isn’t hyperbole.

My Apple Watch Series 5 is running 2 64 bit arm processors, and it’s the same core design that sits on iPhones and iPads from the same generation.

Again, I’m not trying to be “bombastic” or use “hyperbole” but intellectually speaking, people who have ran Atom chips aren’t singing about how great the experience is. The reputation it has was well earned.

I can’t tell you how many times I heard complaints about how bad a netbook was about 10 years ago.

Again, if you want to use terms like “smack” or “hyperbole” that is your right. This isn’t a “pro Apple” or “anything but Intel” thing. I want, desperately for Intel to succeed. I also want AMD, Qualcomm, and NVidia to succeed It makes the market better for everyone.

Congrats for making a generalized statement of “smack”. I don’t know what to tell you.

Intel will sell a lot of “upgraded” Atoms. They’ll be in inexpensive cable modems.

Intel at it’s best is when their on there knees, with their back against the ropes. I want that sharp Intel to come out of the shadows.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,092
1,622
136
You can say what you want, I have no desire to talk “outrageous smack”.

Intel failed in the market place with Atom. That isn’t hyperbole.

My Apple Watch Series 5 is running 2 64 bit arm processors, and it’s the same core design that sits on iPhones and iPads from the same generation.

"found it packs the same punch as the 2015 iPhone model."

And Series 5 has same CPU as Series 4. I agree the CPU is impressive but its like the last generation Atom not Tremont. It also has half the amount of cores.

Also I'm pretty sure you are the one who started the whole derailing of the thread.
 
  • Like
Reactions: coercitiv

coercitiv

Diamond Member
Jan 24, 2014
3,853
3,822
136
Again, I’m not trying to be “bombastic” or use “hyperbole” but intellectually speaking
I agree, bombastic people wouldn't speak intellectually. They would end their diatribe with some Drop The Mic expression like "Prove me wrong" and then continue to write unstructured walls of text since every sentence they produce requires proper spacing to be gazed upon.

I'm so glad we don't have one of these bombastic people around, instead we're just having a civil conversation about Lakefield with a small discussion on the side about the pace of progress (or lack of) in Intel's small cores.
 
  • Like
Reactions: geegee83

mikk

Platinum Member
May 15, 2012
2,826
658
136
Even Lakefield suffers from the 10nm process, they need 10nm SuperFin. If Lakefield-R is based on Gracemont+Willow+Gen12LP it should be much better. Even better would be Golden Cove instead Willow Cove.
 

mikk

Platinum Member
May 15, 2012
2,826
658
136
Alder Lake-P gets at least 2 big cores, Lakefield likely only 1 big core. But of course Lakefield-R might use Willow Cove.
 

Roland00Address

Golden Member
Dec 17, 2008
1,951
77
91
Even Lakefield suffers from the 10nm process, they need 10nm SuperFin. If Lakefield-R is based on Gracemont+Willow+Gen12LP it should be much better. Even better would be Golden Cove instead Willow Cove.
Don't we know what the successor to Lakefeld to be? Isn't it Alder Lake?

Alder Lake being a hybrid chip mixing

Gracemont (next gen "small core" aka Atom cores replacing Tremont Atom Cores.)

Golden Cove (next gen "big core", before Golden Cove is Willow Cove cores which is inside Tiger Lake and Rocket Lake aka Intel's 11th gen family, and before Willow Cove is Sunny Cove which is in Ice Lake and Ice Lake is some of Intel's 10th gen family but not all of the 10th gen.)

We do not know when it will come out? 2021, 2022? Intel per their slide is targeting 2021.

We know it will be a 10nm part.

Ananadtech thinks it will be 8+8, but I personally am skeptical they can put 8 big cores in such a small chip. Here is Dr. Ian Cuttess from anandtech when this chip was announced.

 

mikk

Platinum Member
May 15, 2012
2,826
658
136
ADL-S is up to 8+8 but it's not a Lakefield successor. About Lakefield-R, we do know that it will use Gen12LP graphics. Beside this nothing is known. Thoretical they could scale down ADL-P and cancel Lakefield-R.
 

vigilant007

Junior Member
Dec 7, 2014
21
10
81

"found it packs the same punch as the 2015 iPhone model."

And Series 5 has same CPU as Series 4. I agree the CPU is impressive but its like the last generation Atom not Tremont. It also has half the amount of cores.

Also I'm pretty sure you are the one who started the whole derailing of the thread.
If I derailed the thread, the humbles of apologies, that was far from my intent.

In regards to the CPU, I was speaking to the individual core versus total package. In regards to Tremont, I have looked, though not exhaustively, and haven't seen anything really that indicates conclusively what the performance benefits of it. If memory serves me correctly, the version prior to Tremont was an incremental update.
 

jpiniero

Diamond Member
Oct 1, 2010
8,280
1,355
126
ADL-S is up to 8+8 but it's not a Lakefield successor. About Lakefield-R, we do know that it will use Gen12LP graphics. Beside this nothing is known. Thoretical they could scale down ADL-P and cancel Lakefield-R.
Lakefield Refresh is supposed to be released in early 2021, which is much earlier than the Alder-P timing. Now it could still end up being cancelled.
 

TheGiant

Senior member
Jun 12, 2017
690
294
106
Lakefield Refresh is supposed to be released in early 2021, which is much earlier than the Alder-P timing. Now it could still end up being cancelled.
IMO lakefield R is alder lake P in 1+4, just defect dies produced to make the stock worth 2+N release
 

IntelUser2000

Elite Member
Oct 14, 2003
7,092
1,622
136
If memory serves me correctly, the version prior to Tremont was an incremental update.
The original OoOE Atom was Silvermont. Then we have Airmont which is a 14nm shrink and is essentially the same core perf-wise.

Goldmont is 30-40% faster per clock. Goldmont Plus, which is the predecessor to Tremont is another 30% faster per clock. Let's call 30% incremental.

IMO lakefield R is alder lake P in 1+4, just defect dies produced to make the stock worth 2+N release
Whatever successor to Lakefield that uses Golden Cove and Gracemont is an entirely separate product and die. Alderlake isn't under the obligation to have a die under 100mm2 or C10 idle power be at 2mW. Defect dies suggest disabling cores which means it'll exist physically which will increase die size.
 

IntelUser2000

Elite Member
Oct 14, 2003
7,092
1,622
136
Goldmont Plus was more like 15% better.
Where the hell did you get that?

ST Integer score 2706 vs. 2036

Cinebench ST/MT: 35%
 

ASK THE COMMUNITY