Intel Skylake / Kaby Lake

Page 618 - 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
But this is why I am kinda pissed at Intel. Broadwell C is DDR3 based while Skull Canyon supports fast DDR4. If it had an unlocked TDP on a proper desktop platform, it would annihilate Broadwell-C!

They should support these kind of products, by treating it as a product made by a startup company. I don't think they do, which is why they bail. If it can be called a success on its own, then keep providing them. But nooo, it has to be a venture where it can provide $10 billion revenue in 2 years.
 

nvgpu

Senior member
Sep 12, 2014
629
202
81

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136

TahoeDust

Senior member
Nov 29, 2011
557
404
136
Well, it's on LN2 so is it really astounding?

But is it on LN2?

Obviously i cant read Chinese or whatever language is it in, so i cant read the surrounding texts...

4.5 ghz from 2.6 stock ? That has to be on LN2.

LOL! Yeah, that's not LN2.

Not LN2 confirmed...

Intel Skylake / Kaby Lake / Coffee Lake Thread - Coffee Lake-S specs out (page 554)
yes I just using ek wb now. no delid :)
 

beginner99

Diamond Member
Jun 2, 2009
5,208
1,580
136
Question if someone wanted to run
2 Gpus( I know they would run at 8x8)
2 Nvme ssds
1 Sata ssd
1 optical drive
Would Kaby/Coffee Lake paired with a z270/z370 be able to handle that or would a Skylake X be better suited for the task?

Does sound like you could profit from HEDT at least a bit. Both GPUs could run at 16x speed (yeah not really a huge issue) and you get full speed on ssds. As as been said with Z270/370 you are limited by connection from Chipset to CPU which is 4x pcie 3.0. So only 1 of your nvme ssds can ever run at full speed. So depending on what you do, there could be a bottleneck here. But if you just play games and to some encoding, then I doubt it.

Actually I'm not sure what happens when copying from ssd to ssd, if the data can go through directly or still has to move over DMI. Has this ever been tested? Or is this documented by intel?
 

aigomorla

CPU, Cases&Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
20,839
3,174
126
My X299 Auros Gaming 9 is one hell of blingy board...

2mee1d4.jpg


I want to see some 5ghz+ numbers so i had to get my rack all prep and set...
Was testing the cpu and board before i fully convert it to a torture station.

This is a 7740k... yeah.. i know the gimped processor in the line, but i want to see 5ghz+ numbers, and its only temporary until i can secure a 7980X... which is ultimately why i got the Gaming 9.

LOL u guys can see both extreme's of watercooling on my rack... the left side is about full bloated high tier LCS while the right side is an AIO.... which i love btw, cuz it says my name on the pump.
 
Last edited:

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
yes I just using ek wb now. no delid :)
That;s impressive. That's some premium silicon on that 7980XE. Before your test release, there were some who taught the 7980XE could barely overclock to 4Ghz. You did 4.8Ghz!! Congrats!
 

coffeeblues

Member
Jun 23, 2017
49
18
36
Question if someone wanted to run
2 Gpus( I know they would run at 8x8)
2 Nvme ssds
1 Sata ssd
1 optical drive
Would Kaby/Coffee Lake paired with a z270/z370 be able to handle that or would a Skylake X be better suited for the task?

Unless you want >4GB/s aggregate nvme read or write you can run it with z270/z370 chipset connected ssds, Skylake-X for higher aggregate nvme speeds.

Does sound like you could profit from HEDT at least a bit. Both GPUs could run at 16x speed (yeah not really a huge issue) and you get full speed on ssds. As as been said with Z270/370 you are limited by connection from Chipset to CPU which is 4x pcie 3.0. So only 1 of your nvme ssds can ever run at full speed. So depending on what you do, there could be a bottleneck here. But if you just play games and to some encoding, then I doubt it.

Actually I'm not sure what happens when copying from ssd to ssd, if the data can go through directly or still has to move over DMI. Has this ever been tested? Or is this documented by intel?

DMI/PCIE is bidirectional, so you get almost 4GB/s read and write respectively for nvme to nvme copy scenario. OS handles the drivers, the applications and the access control for device to device communication, devices don't interact directly with each other.
 
Aug 11, 2008
10,451
642
126
That sucks really, really bad. Let me explain.

If they release Cannonlake in mid-2018, there's zero chance we'll see Icelake before mid-2019.

Now that assumes they are still on the 12 months cycle. There's evidence we're past that.

June 2013: Haswell
Jul 2014: Haswell Refresh
August 2015: Skylake
Jan 2017: Kabylake
November 2017?: Coffeelake
Summer 2018: Cannonlake

Hmm. Yea we're not on the 12 month cycle. Based on that pattern,

Fall 2019: Icelake
Holiday 2021: Tigerlake
Spring 2022: 7nm

We are talking 3 year delay for their 10nm process. Basically, their claims of "3.5 year lead" doesn't exist even if you took their(likely marketing) claims at face value.



Proper yield learning likely requires it to be field tested. The best way is to send it to HVM. If they cancel Cannonlake, then Icelake has to bear that burden. That's why when people claim they should "skip" a process, they have zero clue what they are talking about. If you are having a problem on a simpler process, you'll have worse times on a more advanced one.

Ivy Bridge was released in 2012. That means 10 nm should have been released in 2016. If CL is released in 2018, that means a 2 year delay. Icelake is not just a die shrink, but a new architecture as well, so should have been released in 2017. So we are still looking at more like a 2 year delay, although desktop is basically another generation behind for part of that time.
 

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
I still think Ice Lake desktop will be 4q18 at the latest. Intel won't want to leave CFL in the channel any longer than necessary to maintain the component ecosystem and maintain their own upgrade cycle.
 
  • Like
Reactions: Pick2 and Sweepr

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Ivy Bridge was released in 2012. That means 10 nm should have been released in 2016. If CL is released in 2018, that means a 2 year delay. Icelake is not just a die shrink, but a new architecture as well, so should have been released in 2017. So we are still looking at more like a 2 year delay, although desktop is basically another generation behind for part of that time.

You are right. It is two years.

I'm not sure how they'll get Icelake in by end of 2018 if they can get Cannonlake barely by mid 2018. Either we'll see much more at early 2018 or we'll see significant shortcuts taken in performance with Icelake as well.
 

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
You have to remember that Cannonlake was completed a long time ago. The only reason it's not out is because the 10 nm yield is so horrifically bad. Unless Icelake is using EMIB the release is really dependent on if/how much yield improves, and not Cannonlake, esp since Intel seems to be burying it.
 
  • Like
Reactions: Ajay
Mar 10, 2006
11,715
2,012
126
You are right. It is two years.

I'm not sure how they'll get Icelake in by end of 2018 if they can get Cannonlake barely by mid 2018. Either we'll see much more at early 2018 or we'll see significant shortcuts taken in performance with Icelake as well.

Who knows, Intel has long stopped being transparent with the public.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Extremely.

Excessive competitive pressure is cited to be the reason there's less papers to Hot Chips and ISSCC in the recent years.

Unless Icelake is using EMIB the release is really dependent on if/how much yield improves, and not Cannonlake, esp since Intel seems to be burying it.

If the 10nm problems affect performance too,

Then how can we expect to see a 8700K-like clock on a Icelake K part? I think it'll be extremely difficult to do that before 2019.

Problems with having too many defects might be solved by using two 50mm2 dies on MCM(cheapest multi-die solution). Problems with circuit refusing to clock beyond a certain point is a different story.
 

TheLycan

Member
Mar 8, 2017
34
11
36
Excessive competitive pressure is cited to be the reason there's less papers to Hot Chips and ISSCC in the recent years.



If the 10nm problems affect performance too,

Then how can we expect to see a 8700K-like clock on a Icelake K part? I think it'll be extremely difficult to do that before 2019.

Problems with having too many defects might be solved by using two 50mm2 dies on MCM(cheapest multi-die solution). Problems with circuit refusing to clock beyond a certain point is a different story.
Guys there is another thread for icelake, please use that.

Sent from my HTC One using Tapatalk