Intel Skylake / Kaby Lake

Page 512 - 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
There was that Geekbench leak that Intel was testing fused-in LPDDR4X with Icelake that broke the memory latency test. Don't need edram if the system memory is that fast.

That doesn't make sense to me. If they are using HBM2, the latency is not great. HMC used in Knights Landing has higher latency(10-20%) than the 6-channel DDR4. I also believe Icelake is the generation that will bring titanic level of changes in memory hierarchy, but the Geekbench result may be just a fluke.

Just putting memory physically closer does not make it faster. Phones and Tablets all use PoP memory. They are just regular performing system memory.

eDRAM however, is closer to cache than system memory. While the large cell size makes it not suitable for a long term system memory replacement, it's very fast.
 

Malogeek

Golden Member
Mar 5, 2017
1,390
778
136
yaktribe.org
That's 400W from the wall though, right, and for the entre system with GPU? That just doesn't seem like that much to me, especially when most people are running 650-1000W power supplies. I could very well be misinterpreting something though.
Yes that's system power, with GPU at idle. For comparison the Ryzen 1800X at stock during the same test was pulling around 200w system power. Overclocked the 1800X pulls an extra 60ish watts. The 7900X was pulling 150w extra. Crazy.
 

Ranulf

Platinum Member
Jul 18, 2001
2,348
1,165
136
Just for comparison, my 8350 stock 4ghz with a Rx480 uses about 320-400w depending on the game being played.
 

DooKey

Golden Member
Nov 9, 2005
1,811
458
136
I'm leaning towards a CFL 6c/12t instead of a SKL-X because I want more cores but better gaming than my current 6700K. Can't wait to see some benchmarks.
 
  • Like
Reactions: TheF34RChannel

TheF34RChannel

Senior member
May 18, 2017
786
309
136
Intel has all the motivation and no alternatives but to succeed in clocking it higher than r5 1600x , so the end result should be that good.

I am kinda hoping for a 4.5GHz boost on all cores with some room left for user improvements on those clocks. Should be doable @ 14nm++

I'm leaning towards a CFL 6c/12t instead of a SKL-X because I want more cores but better gaming than my current 6700K. Can't wait to see some benchmarks.

Sounds right and unlikely that you will be disappointed when playing your games that know what to do with the additional two cores.
 
  • Like
Reactions: DooKey

Saylick

Diamond Member
Sep 10, 2012
3,125
6,296
136
I am kinda hoping for a 4.5GHz boost on all cores with some room left for user improvements on those clocks. Should be doable @ 14nm++



Sounds right and unlikely that you will be disappointed when playing your games that know what to do with the additional two cores.
I think I am in the same boat. 14nm++ should offer another 10% increase in clocks so I think 4.5 GHz flat on 6C is definitely possible for Coffeelake. Ideal scenario would be 5 GHz on all cores.

In terms of future proofing, I think 6C may be the sweet spot for the next 5 years. 4C is already showing age and modern consoles really only use 6-7 cores during gaming (with 1-2 being dedicated for the OS and background tasks). 8C would definitely be a long term play but I think a modern 8C would be outdated by the time 8C becomes the new 4C.
 

TheF34RChannel

Senior member
May 18, 2017
786
309
136
I think I am in the same boat. 14nm++ should offer another 10% increase in clocks so I think 4.5 GHz flat on 6C is definitely possible for Coffeelake. Ideal scenario would be 5 GHz on all cores.

In terms of future proofing, I think 6C may be the sweet spot for the next 5 years. 4C is already showing age and modern consoles really only use 6-7 cores during gaming (with 1-2 being dedicated for the OS and background tasks). 8C would definitely be a long term play but I think a modern 8C would be outdated by the time 8C becomes the new 4C.

We're probably way beyond Tiger Lake when we find 8C as the standard for gaming.

As for CFL-S stock boost clocks, I'm looking for a lazy OC by just clicking boost on all cores in the BIOS lol, and with 7700 clocks being a possibility on the new 6C that'd be an ideal 4.5 which is all I really need, and if not I can always put my own OC on easy enough.
 

RichUK

Lifer
Feb 14, 2005
10,320
672
126
Has anyone bought Skylake-X cpu / x299 motherbaord yet?

I can see the CPUs are in stock here, but motherboards are on pre-order.
 

Ajay

Lifer
Jan 8, 2001
15,430
7,849
136
Has anyone bought Skylake-X cpu / x299 motherbaord yet?

I can see the CPUs are in stock here, but motherboards are on pre-order.
I think a couple ppl have, in the Skylake-X thread. They are still waiting on Motherboards.
 
  • Like
Reactions: Pick2

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
1 out of 4 ain't bad!

Sorry, was pretty funny :)
Naa. I find the price okey for bf1, that is the benchmark in this world :). I could use a 7820 because of stellar perf but tdp is over the top for my closet usage with d15 and imo efficiency is a bit old school and a huge disapointment. 10/7nm is where it is for me aparently. But interesting non the less. Still looking how it pans out. Might go and undervolt the sklx a bit. Good times.
 

Malogeek

Golden Member
Mar 5, 2017
1,390
778
136
yaktribe.org
Naa. I find the price okey for bf1, that is the benchmark in this world :). I could use a 7820 because of stellar perf but tdp is over the top for my closet usage with d15 and imo efficiency is a bit old school and a huge disapointment. 10/7nm is where it is for me aparently. But interesting non the less. Still looking how it pans out. Might go and undervolt the sklx a bit. Good times.
Any benchmarks done for BF1 by these reviewers are useless though, as they just use the canned benchmark. However, any of these 6 core+ will be great for BF1. Even better than Ryzen with higher clock speed and lots of cores.
 
  • Like
Reactions: Drazick

imported_ats

Senior member
Mar 21, 2008
422
63
86
New silicon takes much more time than MCM or EMIB does. Both the P4 and C2Q we're reactionary MCM solutions. You can come up with an MCM solution in a year. You have at least 2 year reset on any development if make major changes to silicon.

Modern chips and designs aren't at all comparable to P4 or C2Q. In both cases, they were designed with a multi-drop FSB, which makes sticking 2 in a package pretty trivial. No modern design uses a multi-drop FSB for basic performance reasons.

Adding another slice into a ring is pretty straight forward and can be done pretty much post layout. Intel hasn't been designing 3 separate chips for Xeon these past 5+ year. Instead they've just taken out slices and changed some config wires, it is all 1 layout. Turnaround for adding an unplanned slice into a ring is basically 1 year from decision to PRQ, since it can be done post layout.
 
  • Like
Reactions: Phynaz

TheGiant

Senior member
Jun 12, 2017
748
353
106
Techspot published their review

It is in line with others with finally some more game testing

It looks to me that skl-x is a ryzen style CPU. gaming performance is going up with cores as with ryzen but the baseline with 6C 7800X is just slower than 7700k, that is a dissapointment. I hope coffee lake will do, but I fear there will be some power constrains with 1151 socket. 14++ isn't going to bring less power just more freq headroom....

One chart is inconsistent with what we have seen before, that is power draw

Power2.png

there is clear difference between 6C, 8C and 10C SKL-X, where AnandTech and others measured the same power draw

why is that?
 
  • Like
Reactions: multippp

coercitiv

Diamond Member
Jan 24, 2014
6,187
11,858
136
It looks to me that skl-x is a ryzen style CPU. gaming performance is going up with cores as with ryzen but the baseline with 6C 7800X is just slower than 7700k, that is a dissapointment.
The 7800X comes clocked bellow 4Ghz in stock config (Intel needs their segmentation). I would hold complete judgement until reviewers have the time to overclock it as well and test gaming performance.

To further muddy the waters, even Kaby Lake X managed to get lower gaming results than 7700K, simply because this rushed launch put enough pressure on mobo manufacturers as to ship reviewers units with poorly implemented Turbo schemes.

It's a premium mess.

One chart is inconsistent with what we have seen before, that is power draw

there is clear difference between 6C, 8C and 10C SKL-X, where AnandTech and others measured the same power draw

why is that?
Ananadtech measured power draw under a different load, in Prime95 all SKL-X SKUs are bound to hit 140W and drop frequencies as much as needed to maintain the 140W TDP ceiling. You would see different power numbers in Prime 95 as well if all Skylake X SKUs would be compared at the same clock speed.

When running CB15 power usage is lower, and 7800X with fewer cores and lower max clocks will definitely consume less power.

And what is the representative value? prime or cinebench ?
Both :)