Discussion Intel current and future Lakes & Rapids thread

Page 147 - 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
I don't see how a guaranteed GHz is "largely irrelevant". Yes, the 1.3GHz will certainly turbo as much as it can which is good, but it could fall all the wall back to 1.3GHz vs 2.3GHz. I mean, it's pretty bad when you can't even compare the two chips.

How Icelake will fare in the desktop environment with much higher voltage and TDP is different from this discussion of why the 28W doesn't exist.

And it's not very hard to find laptops that go below the Base frequency when under enough stress.
 
Last edited:

Thunder 57

Platinum Member
Aug 19, 2007
2,647
3,706
136
I explained it in my post. You just seem determined to find something bad to say about intel's products.

Not really. Unlike you, I want the only two x86 companies to compete so we all win. Ice Lake looks good in low power, but can it scale up? Also, nobody here can take you seriously because of your name.

How Icelake will fare in the desktop environment with much higher voltage and TDP is different from this discussion of why the 28W doesn't exist.

And it's not very hard to find laptops that go below the Base frequency when under enough stress.

We shall see. It will be interesting for sure. I don't have a lot of faith in Intel ATM but they have pulled off some amazing stuff before (Conroe).
 

Adonisds

Member
Oct 27, 2019
98
33
51
Not really. Unlike you, I want the only two x86 companies to compete so we all win. Ice Lake looks good in low power, but can it scale up? Also, nobody here can take you seriously because of your name.



We shall see. It will be interesting for sure. I don't have a lot of faith in Intel ATM but they have pulled off some amazing stuff before (Conroe).
I can't take seriously people who don't take someone seriously because of their name.
 
  • Like
Reactions: pcp7

Adonisds

Member
Oct 27, 2019
98
33
51
Sunnycove is originally a 14nm core with a very special style of design. Thus, it is as simple as recompiling its 272-nm cell design to the 340-nm cell design.
Originally 14nm? Could you link the source please?
How did the 10nm and 7nm delays affect the microarchitecture plans? Has the 10nm microarchitecture been ready for years and all this time it had been waiting the process to be ready or have they been redesigning it as the process got delayed? What does the microarchitecture team use the bunch of extra time they got mostly on?
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
Originally 14nm? Could you link the source please?
Sunnycove is/was the in-node successor to Skylake cpu micro-architecture.
Willowcove is/was the in-node successor to Cannonlake cpu micro-architecture.
How did the 10nm and 7nm delays affect the microarchitecture plans? Has the 10nm microarchitecture been ready for years and all this time it had been waiting the process to be ready or have they been redesigning it as the process got delayed?
The original plan before the big mix up was:
Sunnycove on 14nm Hyperscaling and Willowcove on 10nm Hyperscaling.
However, they swapped it for Cannonlake(408CH) to Palmcove(272CH) and Sunnycove(399CH-14nm) to Sunnycove(272CH-10nm). SunnycoveX will be using the 340CH or/and 408CH library for increased performance.

10nm w/o Hyperscaling was good for 2016. "12.92"T Skylake-14nm to "9.27"T Cannonlake-10nm which the 8-core was planned to launch in 2016. Up till the hyperscaling plan was implemented. Which is a far shift from what actually came out "6.18"T Palmcove-10nm. For example, "12.92"T Skylake-14nm to "7.67"T Sunnycove-14nm is the translation for 14nm.

Smaller cell height = lower power and area, plus lower frequency scaling.
Larger cell height = higher performance and leakage, plus higher frequency scaling.
 
Last edited:

Thunder 57

Platinum Member
Aug 19, 2007
2,647
3,706
136
I can't take seriously people who don't take someone seriously because of their name.

Read it backwards and look at his post history if you want. If you still don't take me seriously, well that's fine. I just want to make sure you understand why before making a judgement.

I see you're new here. I wouldn't hold your breath waiting on that if I were you.

I didn't know whether to like this post or find it funny. Too bad I can only choose one.
 

Abwx

Lifer
Apr 2, 2011
10,847
3,297
136
. Ice Lake looks good in low power, but can it scale up? ).

Actually it s not good at all, it score 480 pts@15W in Cinebench R15, FTR at a constant 15W KBL/Raven Ridge/Picasso are at 460/470/500 pts respectively.

 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
Actually it s not good at all, it score 480 pts@15W in Cinebench R15, FTR at a constant 15W KBL/Raven Ridge/Picasso are at 460/470/500 pts respectively.

Is this what you're talking about?

Image 21.png
 

Abwx

Lifer
Apr 2, 2011
10,847
3,297
136
Is this what you're talking about?

View attachment 13165

Among others...

Now if the score is 830 pts FI this mean that frequency is 830/480 = 1.73x higher and power will be at least 3x higher, like in the case below where the first run is done at 45W :


At this point this CPU perf/watt is easily characterised with the available tests, and laws of physics being what they are there s no need to elaborate further...
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
According to your link, we're looking at 521 ~ 830 points, for an average of 684 points?
 

Attachments

  • Image 211.png
    Image 211.png
    663.9 KB · Views: 14

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
According to your link, we're looking at 521 ~ 830 points, for an average of 684 points?

I highly doubt the differences are big as ABWX suggests. ~600 range needs Icelake to be at 25W, and <500 in the 15W.

But the AMD devices are not using little power as he thinks either. If you look at their power consumption numbers, the 15W chips are in the 30-33W range and the 25W chips are in the 40-45W range. Ignore the Load max and compare Load average because that'll be the time when it stabilizes. That's true for both AMD/Intel. The 40-45W Laverage laptops do better than the 30-33W range laptops.

The utilities running on AMD are underreporting chip power usage by quite a bit. One 3750H laptop shows 15-22W. Seriously.

If you compare the Surface Laptop 3 with Ryzen 5 against Surface Laptop 2 with 8250U, the former is in the 600+ range and its Laverage is at 45W. The SL2's Laverage is at 35W, or nearly 10W lower. SL2's review says the CPU eventually stabilizes at 18W. So that 10W approximates the sustained chip power differences of 25W vs 15W.

Yet the HWInfo64 utility reports 9W average for the Ryzen 5.
 
Last edited:

eddman

Senior member
Dec 28, 2010
239
87
101
The original plan before the big mix up was:
Sunnycove on 14nm Hyperscaling and Willowcove on 10nm Hyperscaling.
However, they swapped it for Cannonlake(408CH) to Palmcove(272CH) and Sunnycove(399CH-14nm) to Sunnycove(272CH-10nm). SunnycoveX will be using the 340CH or/and 408CH library for increased performance.

10nm w/o Hyperscaling was good for 2016. "12.92"T Skylake-14nm to "9.27"T Cannonlake-10nm which the 8-core was planned to launch in 2016. Up till the hyperscaling plan was implemented. Which is a far shift from what actually came out "6.18"T Palmcove-10nm. For example, "12.92"T Skylake-14nm to "7.67"T Sunnycove-14nm is the translation for 14nm.
I'm not a regular here. Where do you get these information from, library codes, etc? Are you an insider?
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
I'm not a regular here. Where do you get these information from, library codes, etc? Are you an insider?
IEDM, IEEE, Intel slides, Linkedin leaks.
He is Full of bs, just ask where tunnelborer is and how cmt is the preferred Amd design not zen.
Willowcove = Tunnelborer
Goldencove = Digger
Tremont = Crane
Gracemont = Harvester

Same architects = same architecture, while Intel is clustered, AMD is struggling with enlarged single cluster SMT. If there was ever a phrase to come to terms with what happened: "Omae wa mou shindeiru" , is the best.

AMD herp-derp here is our architecture that is just like our 5 GHz K9 architecture.(b-June 2007, rc-January 2010, augmented open64)
Intel oh, here we are launching what you didn't dare.

1. Tigerlake-S
2. Alderlake-S
Punches, and they are out for good.
 
Last edited:

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
IEDM, IEEE, Intel slides, Linkedin leaks.
Willowcove = Tunnelborer
Goldencove = Digger
Tremont = Crane
Gracemont = Harvester

Same architects = same architecture, while Intel is clustered, AMD is struggling with enlarged single cluster SMT. If there was ever a phrase to come to terms with what happened: "Omae wa mou shindeiru" , is the best.

AMD herp-derp here is our architecture that is just like our 5 GHz K9 architecture.(b-June 2007, rc-January 2010, augmented open64)
Intel oh, here we are launching what you didn't dare.

1. Tigerlake-S
2. Alderlake-S
Punches, and they are out for good.

This is, as usual, 100% nonsense. First you claimed that Tunnelborer (or whatever the lunacy of the day is) was a FDSOI CMT chip, and now you say it's a completely different core from another company entirely? Not even from the same continent!

And why don't you post these "IEDM, IEEE, Intel slides, Linkedin leaks" then, hmm?
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,683
1,218
136
This is, as usual, 100% nonsense. First you claimed that Tunnelborer (or whatever the lunacy of the day is) was a FDSOI CMT chip, and now you say it's a completely different core from another company entirely? Not even from the same continent!
The architects left AMD to go SoftMachines(~40%) or Intel(~50%). I also stated AMD would use Mariana trench or something, hint Oceancove = AMD's Mariana core. Zen successors shoot for the stars(born to fail), Bulldozer successors stayed grounded on Earth(born to succeed). This is a statement not by me, but was posted in a leak, after someone left AMD.
And why don't you post these "IEDM, IEEE, Intel slides, Linkedin leaks" then, hmm?
They are easily found. No effort for me, equals no effort for you.

Folsom cores are better than anything on the market. Tigerlake's WLC will prove it.
 
Last edited:
  • Like
Reactions: majord

eddman

Senior member
Dec 28, 2010
239
87
101
This is a statement not by me, but was posted in a leak, after someone left AMD. They are easily found. No effort for me, equals no effort for you.
Then why won't you post the links? When Dayman1225 claims something, he posts or links the actual information for everyone to see.

You make grand claims and yet provide nothing. Why? If they are so easily found, then be kind and share. Not all of us have the skills or the knowledge to find such intricate information.

"The burden of proof is upon the claimant."
 

Thunder 57

Platinum Member
Aug 19, 2007
2,647
3,706
136
The architects left AMD to go SoftMachines(~40%) or Intel(~50%). I also stated AMD would use Mariana trench or something, hint Oceancove = AMD's Mariana core. Zen successors shoot for the stars(born to fail), Bulldozer successors stayed grounded on Earth(born to succeed). This is a statement not by me, but was posted in a leak, after someone left AMD.They are easily found. No effort for me, equals no effort for you.

Folsom cores are better than anything on the market. Tigerlake's WLC will prove it.

"They are easily found". Translation: "They don't exist". It takes maybe a minute to link something. If you want to be credible, you would do that.
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
Relax guys enjoy. Nosta is a good creative guy. Its not each day you have to read it. Let it be.

The problem is that you have new, infrequent, or even just gullible users who don't know that he only posts nonsensical technobabble. Imo, if someone wants to post their hardware fan fiction without any disclaimer about its nature, they should keep it to a personal blog and off actual tech forums.
 
Last edited:

Thunder 57

Platinum Member
Aug 19, 2007
2,647
3,706
136
The problem is that you have new, infrequent, or even outright gullible users who don't know that he only posts nonsensical technobabble. Imo, if someone wants to post their hardware fan fiction without any disclaim about its nature, they should keep it to a personal blog and off actual tech forums.

It is frustrating and not at all beneficial to the forums. I'm all for speculation but posting nonsense as fact is annoying.
 
  • Like
Reactions: Gideon