Intel Skylake / Kaby Lake

Page 631 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

dullard

Elite Member
May 21, 2001
24,998
3,325
126
In the same graph, the power reduction (at same performance point 1.0 in X axis) from 14nm (Core i7 6700K 4C 8T) to 14nm+ (Core i7 7700K 4C 8T) is higher than 14nm+ (Core i7 7700K 4C 8T) to 14nm++ (Core i7 8700K 6T 12T). So if Core i7 7700K with the same number of cores vs 6700K and same TDP only got 300MHz higher turbo, then its illogical to expect same turbo as 7700K with 50% more Cores/Threads from 14nm++.
The illogical portion of your post is to relate TDP and turbo speed without any other data.

For example, Intel could put the turbo speed of the 8700K astronomically high (lets say, 10 GHz) if it wanted to spend nearly 0% of the time at full turbo. Or maybe it could set it to 6 GHz, if it only spent 1% of the time at full turbo. Or maybe 5 GHz if it spends 80% of the time at full turbo, etc (all numbers in this paragraph are just for discussion and are not based on benchmarks).

Thus, the turbo speed, without knowledge of the amount of time spent at full turbo is a near meaningless number. It is thus illogical to do what you are attempting to do, which is relate these two numbers without the third critical number (percent of time possible to stay at full turbo when the cooling system TDP is at the TDP of the processor).
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
KL actually has a 400mhz higher all core turbo than SL, and a 300mhz higher single core turbo than SL.
That's a big difference.

Yes but they both 4C 8T, KL to CFL is 50% more cores/threads

Its easier to compare 7800X which is 6C 12T at 14nm+ against CFL, if CFL at 95W TDP reach the same MT performance as 7800X at 140W TDP then it should be more than fine and within what we would expect from 14nm+ to 14nm++.
 

LTC8K6

Lifer
Mar 10, 2004
28,520
1,575
126
Yes but they both 4C 8T, KL to CFL is 50% more cores/threads

Its easier to compare 7800X which is 6C 12T at 14nm+ against CFL, if CFL at 95W TDP reach the same MT performance as 7800X at 140W TDP then it should be more than fine and within what we would expect from 14nm+ to 14nm++.
But it's a lower all-core turbo for the 8700K than the 7700K.
7800X has a lot more L2 cache than 7700K
7740X has a higher all core turbo than 7700K
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
But it's a lower all-core turbo for the 8700K than the 7700K.
7800X has a lot more L2 cache than 7700K
7740X has a higher all core turbo than 7700K

Yes, 7740X has higher turbo but it also has 112W TDP vs 91W TDP of the 7700K and both have the same number of Cores/Threads.
7800X has 47% higher TDP thermal buffer and larger die, it will maintain the highest all core turbo much longer than 6C 12T CFL.
So when im saying that if CFL can reach the 7800X MT performance is to be seen as a good thing and people shouldn't expecting miracles. The strength of the 8700K will be its ability to OC a 6C 12T CPU, not its default performance at 95W TDP.
 

ezodagrom

Junior Member
Mar 4, 2009
7
11
81
Yes but they both 4C 8T, KL to CFL is 50% more cores/threads

Its easier to compare 7800X which is 6C 12T at 14nm+ against CFL, if CFL at 95W TDP reach the same MT performance as 7800X at 140W TDP then it should be more than fine and within what we would expect from 14nm+ to 14nm++.
7800X isn't the only one with a 140W TDP though.
The higher clocked 8c/16t 7820X has a 140W TDP as well. The 10c/20t 7900X and 12c/24t 7920X also have a 140W TDP, but these are lower clocked than the 7800X/7820X (though the 7900X has the same turbo as the 7820X).
 
  • Like
Reactions: eddman

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
7800X isn't the only one with a 140W TDP though.
The higher clocked 8c/16t 7820X has a 140W TDP as well. The 10c/20t 7900X and 12c/24t 7920X also have a 140W TDP, but these are lower clocked than the 7800X/7820X (though the 7900X has the same turbo as the 7820X).

We take the 7800X as a reference because it has the same number of Cores/Threads as the 8700K.
 

dullard

Elite Member
May 21, 2001
24,998
3,325
126
We take the 7800X as a reference because it has the same number of Cores/Threads as the 8700K.
That is a fairly bad reason to use it. The number of cores usually ends up as being one of the worst metrics to hold constant if you want good comparison data.

The 7800K is the runt of the Skylake-X litter. The 8700K is the best of Coffee Lake.

TDP is also quite arbitrary. Intel could slap on a higher number than is really needed to ensure 7800K users have an upgrade path or to ensure that the yield is high where all bad Skylake-X chips can at least be sold as 7800K chips.
 
Last edited:

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
7700k 4C/8T - 4.2 Ghz base, 4.4 Ghz All core turbo, 4.5 Ghz max turbo.
8700k 6C/12T - 3.7 Ghz base, 4.3 Ghz All core turbo, 4.7 Ghz max turbo.

I am with atenra in having doubts if with just 10-12% higher transistor performance (14++ vs 14+) Intel can fit 50% more cores into roughly the same TDP with just 100 Mhz loss of all core turbo. I am waiting to see how well 8700k maintains/sustains its all core turbo in workloads like CB R15 especially those which are of longer duration. I think 8700k will be a beast of an overclocker.
 
Last edited:

dullard

Elite Member
May 21, 2001
24,998
3,325
126
7700k 4C/8T - 4.2 Ghz base, 4.4 Ghz All core turbo, 4.5 Ghz max turbo.
8700k 6C/12T - 3.7 Ghz base, 4.3 Ghz All core turbo, 4.7 Ghz max turbo.

I am with atenra in having doubts if with just 10-12% higher transistor performance (14++ vs 14+) Intel can fit 50% mores into roughly the same TDP with just 100 Mhz loss of all core turbo. I am waiting to see how well 8700k maintains/sustains its all core turbo in workloads like CB R15 especially those which are of longer duration. I think 8700k will be a beast of an overclocker.
I just posted Intel's data of 27% lower power for the same transistor performance. Why are you using 10 to 12%?

I too have doubts about sustaining the high turbos that have been rumored. But at least, start from the one known data set that we have from Intel in your analysis.
 

Lodix

Senior member
Jun 24, 2016
340
116
116
Just 100MHz clock speed reduction at that high frequencies could mean a big drop in power consuption ( anyone has a frequency/power curve for SK or BL ? ). Add the new process and refinements on the architecture/desing, plus a bit bigger die to dissipate heat... I don't see anything crazy with the rumoured clocks for the 8700k.
 
Mar 10, 2006
11,715
2,012
126
7700k 4C/8T - 4.2 Ghz base, 4.4 Ghz All core turbo, 4.5 Ghz max turbo.
8700k 6C/12T - 3.7 Ghz base, 4.3 Ghz All core turbo, 4.7 Ghz max turbo.

I am with atenra in having doubts if with just 10-12% higher transistor performance (14++ vs 14+) Intel can fit 50% mores into roughly the same TDP with just 100 Mhz loss of all core turbo. I am waiting to see how well 8700k maintains/sustains its all core turbo in workloads like CB R15 especially those which are of longer duration. I think 8700k will be a beast of an overclocker.

Sustained frequency on desktop should be more a function of your cooling solution than anything else. Mobile (i.e. CFL-H) on the other hand is a different ballgame and will be where the rubber hits the road for 14nm++'s efficiency improvements.
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
7800X has a different cache structure, quad channel vs dual channel, has AVX-512 support, and designed for use in a totally different platform.
This is the missing piece right here. Skylake X TDPs are based on this, from what I've observed. That's why you can't compare coffee lake with SKL-X.
 
  • Like
Reactions: Arachnotronic

mikk

Diamond Member
May 15, 2012
4,111
2,105
136
I think that the turbos are working as intended, the ST turbo on the leaked 196 ST score and 1410 score on MT are both accurate enough for a stock configuration.

196 ST corresponds to Kabylake with 4.5 Ghz. You are saying 4.5 Ghz ST is accurate for 8700k, I say it is flawed because ST Turbo runs with 4.7 Ghz on this SKU. This is clearly not a representative system. You are wrong with your posted MT score by the way.
 
  • Like
Reactions: IEC

dwade

Junior Member
Jun 25, 2017
19
24
36
Let the red team use it to troll though. They won't have much ammo left to use after October 5th besides the usual "good enough," "but it costs less," "I get to upgrade to Zen2 on the same board," "Intel and Nvidia are evil. AMD are the good guys."
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,323
4,904
136
Anyone who believes the absurdly low CB score is with a working turbo... lol

It will beat the 1600X in MT. No doubt.
 
  • Like
Reactions: pcp7

jpiniero

Lifer
Oct 1, 2010
14,509
5,159
136
Damn, you know how unpopular Skylake-R was? Intel started the EOL process on it in December. It was released in May. I'm not sure they actually got any design wins.
 

FIVR

Diamond Member
Jun 1, 2016
3,753
911
106
It will beat the 1600X in MT. No doubt.

Guru3d seems to think it's legit.

http://www.guru3d.com/news-story/intel-core-i7-8700k-benchmarks.html

I have my doubts that Intel would bother releasing a chip that couldn't compete core-for-core against AMD's lower end 6 core in MT. However, we've already seen how Ryzen has better MT scaling than Intel and we've also already seen that Ryzen uses less power. I think it's possible intel squeezed everything it could out of the skylake architecture and couldn't get the chip to compete in MT anyway.
 

The Stilt

Golden Member
Dec 5, 2015
1,709
3,057
106
Ryzen is extremely hard to beat in Cinebench R11.5 / R15 MT due to it's significantly (some might say unrealistically) high(er) SMT yield.
In Cinebench R15 KBL has SMT yield of 30.4% while Ryzen has 41.6%. On average the difference is much smaller, ~25.4% for KBL and 28.7% for Ryzen.

Regardless, the claimed score of 196 in CB15 at 3.7GHz is nothing but "CUBS". Unless of course CFL has ~21% IPC advantage over KBL (which it doesn't).
 
  • Like
Reactions: Drazick