Intel Skylake / Kaby Lake

Page 149 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Dave2150

Senior member
Jan 20, 2015
639
178
116
So in resume, Skylake is FAR better than SB in terms to getting up to 4.7-4.8 Ghz without adding voltage, but going further gets quickly worse?

Also.. someone did put the chip up to 5.2 Ghz STABLE? I remember that SB is capable to do that (with water cooling)

The TIM is the problem for 4.8-5Ghz. Temperatures just skyrocket at 1.4v, even under a h110i gt.

I'd imagine Skylake-E will be an absolute beast of a clocker/performer with a soldered die.
 

Dave2150

Senior member
Jan 20, 2015
639
178
116
Wow, I just did something I haven't done in a while...buy a CPU/MB combo from somewhere OTHER than Microcenter.

$566 shipped for the 6600k , GA-Z170X, and 16GB G.skill memory, not really anymore than it would have cost from MC and I don;t have to drive an hour+ to get it.

Looks like it's time to move the venerable 2500k to the htpc :)

Should be a very nice system :)
 

sxr7171

Diamond Member
Jun 21, 2002
5,079
40
91
They were available on August 5th in the UK, and they are still in stock today.

Had my 6700k and z170 motherboard delivered on the 6th of August:

f6x3otZ.jpg

Settled on 4.7Ghz with cache at 4.4Ghz for my 24/7 overclock, at 1.29v. Runs very cool and quiet at this speed:

CO9u30x.png


I can bench at 5Ghz, though temperatures get too high for everyday use. I could delid it and run at this speed, though that's a step too far for my comfort zone :p

Wow those are excellent results
 

CHADBOGA

Platinum Member
Mar 31, 2009
2,135
832
136
Wow, I just did something I haven't done in a while...buy a CPU/MB combo from somewhere OTHER than Microcenter.

$566 shipped for the 6600k , GA-Z170X, and 16GB G.skill memory, not really anymore than it would have cost from MC and I don;t have to drive an hour+ to get it.

Looks like it's time to move the venerable 2500k to the htpc :)
What speed ram?
 

myocardia

Diamond Member
Jun 21, 2003
9,291
30
91
What does it say? Claiming there is zero difference is false. Not all applications are latency insensitive.
So, would you like to guess much this 45.8% DDR3 latency advantage gains the 6700k? Zero percent (or extremely close to it), compared to the 2,133 DDR4:
7gvjrbty.png

It says that there is a 0% difference, or close enough to zero not to matter, just as your example proves. Or are you actually claiming that the widest margin shown in your cherry-picked example, 3.17%, is huge, and is a much larger percentage than the 45.8% latency advantage? Thanks for proving my point.
 
Last edited:

mikk

Diamond Member
May 15, 2012
4,112
2,108
136
It says that there is a 0% difference, or close enough to zero not to matter, just as your example proves.


I remind you:

More important yes, I wouldn't say much more important. DDR4-2133 CL15 isn't necessarily better than DDR3-1600 CL9.



This example proves me! I told DDR4-2133 CL15 isn't necessarily better than DDR3-1600 CL9 and here we have an example where DDR3-1600 CL9 is even better than DDR4-2133 CL15.

According to you this shouldn't happen. Furthermore DDR3-2133 CL11 is 12.5% faster than DDR4-2133 CL15 in this case. In fact you need DDR4-2800 CL15-16 to match DDR3-2133 CL11 in this case.

As I said claiming that there is zero difference from better latency and all what matters is bandwidth is plain wrong and false. As said bandwidth is probably more important in most cases but latency isn't something we can ignore.
 
Last edited:

myocardia

Diamond Member
Jun 21, 2003
9,291
30
91
I wasn't, but you didn't show anything. There's plenty of results attesting the benefits of faster memory with lower latency for Skylake and reviews using different memory configurations (even DDR4-2133 CL15) where Skylake compares more favourably to Haswell than AnandTech's tests.
Really? You mean that two different reviews, performed by different people, using different hardware, at difference times, could have a few percentage points of performance difference? Imagine that! :rolleyes:

May I also ask, since you trust AnandTech's results so much, if lower latency DDR3 basically matches faster DDR4 here then why reviews like PCLab's using DDR4-2666 CL16 for Skylake and DDR3-1866 CL9 for Haswell are (according to you) pro-Skylake? Oh oh, looks like you just shot yourself in the foot.
You honestly need to educate yourself about DRAM, or else you will continue to make yourself look like the fool that you have continually made yourself look in this thread. In your example cited above, the person who wrote the unlinked article at PCLab to which you refer is using RAM that is 25.0% higher bandwidth than Ian used in the AnandTech Skylake review: http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation

If you had the ability to comprehend the things posted in this thread not posted by you, specifically this post: http://forums.anandtech.com/showpost.php?p=37632136&postcount=3688 you'd realize that 25% higher bandwidth, when using the same architecture of DRAM, gives you 25% lower latency. If you use RAM with the identical CAS latency, then you get 25.00% lower latency. If you use RAM with a tiny bit higher CAS latency, such as the DDR4 2,666 CL 16 you mention in the unlinked PCLab article, you lose a total of ¾ of a nanosecond of latency, making the 2,666 CL 16 "only" 93.75% as fast, and you "only" end up gaining 23.4375% less latency.

Here are the calculations I used from the above paragraph:
DDDR3 1,866 CL9= 9.64ns of latency
DDR4 2,133 CL 15= 14.06 ns of latency
DDR4 2,666 CL 15= 11.25 ns of latency
DDR4 2,666 CL 16= 12.00 ns of latency

AnandTech's method:
PCLab's review
DDR4-2666/16 = 166 (Skylake-S)
DDR3-1866/9 = 207 (Haswell)
You continue to prove me right. Why do you continue to embarrass yourself? In this example, you've added 23.75% bandwidth over the DDR4 used in Ian Cutress' Skylake review, while lowering the latency by 23.75% (both only for your example), yet you can't figure out how it could perform better? Seriously? Wow.

PCLab's overall gaming performance chart - 14 games @ 1080p
gry.png


But but but how is Skylake beating Haswell by ~13% if, according to AnandTech's methodology (since you trust them with their tests) and Crucial's real latency calculation they are still benefiting Haswell with the memory choice and you just posted tests above where lower latency DDR3 can keep up with faster DDR4?
Your answer to this inaccurate FUD is in the paragraphs above, if you have the capability to comprehend.

I won't waste my time arguing with a butthurt Haswell user pissed off because its toy is now outdated, especially one that labels all tests that don't fit his agenda as 'payed by Intel'.

That's fine by me, but I'm going to continue to embarrass you each and every time you post FUD, like you have been for quite a few pages now.
 

myocardia

Diamond Member
Jun 21, 2003
9,291
30
91
This example proves me! I told DDR4-2133 CL15 isn't necessarily better than DDR3-1600 CL9 and here we have an example where DDR3-1600 CL9 is even better than DDR4-2133 CL15.
Yes, by the utterly, outrageously huge percentage of 3.12%, while using the only software that any normal person on Earth would use while at home, ever. (Edit: while at home, and not doing work) BTW, that 3.12% is gained with the DDR3 having a 45.8% latency advantage, remember.

Furthermore DDR3-2133 CL11 is 12.5% faster than DDR4-2133 CL15 in this case. In fact you need DDR4-2800 CL15-16 to match DDR3-2133 CL11 in this case.
Wait, now we're cherry-picking from cherry-picked examples? Fine, I can do that as well. DDR3 2,666 CL 11 and DDR4 3,600 CL 17 have a 99.7% performance differential, and that's in your cherry-picked software. What now?

As I said claiming that there is zero difference from better latency and all what matters is bandwidth is plain wrong and false. As said bandwidth is probably more important in most cases but latency isn't something we can ignore.
Talk about a strawman. Here is what you were attempting to refute:

myocardia said:
So, would you like to guess much this 45.8% DDR3 latency advantage gains the 6700k? Zero percent (or extremely close to it)


Even the most pedantic person I've ever met would conclude that a maximum of 3.1% difference, when comparing like speeds of RAM from different architectures (2,133 DDR4 vs 1,866 DDR3, 2,800 DDR4 vs 2,133 DDR3, DD4 3,600 vs DDR3 2,666, et al) and a minimum of .3% difference, are in fact in the zero percent to extremely close to zero percent range.
 
Last edited:

Fjodor2001

Diamond Member
Feb 6, 2010
3,705
182
106
http://www.kitguru.net/components/c...al-shortage-of-intel-skylake-microprocessors/

Asustek and Intel confirm global shortage of Intel “Skylake” chips

Chief executive officer of Asustek Computer confirmed on Friday that there is a shortage of Intel Corp.’s “Skylake” processors in the market. Last month a Sweden-based retailer warned enthusiasts about short supply of Intel’s new central processing units with unlocked multiplier. As it appears, the problem is not limited to desktop chips, but also concerns mobile CPUs. Intel admits that there is a shortage of new processors and vows to increase availability.

Apparently, the ramp of certain Intel’s microprocessors produced using 14nm process technology will be rather slow in the third quarter of the year. The reasons for limited supplies and sluggish ramp up are unknown, but not all Intel Core “Skylake” models with unlocked multiplier are available everywhere in the U.S. and Europe*. Moreover, Intel Core i7-5775C and Core i5-5665C “Broadwell” are not even offered by numerous major online stores**.
[...]
Intel admitted that the “Skylake” processors are in short supply and promised to improve availability later in the third quarter.

“We are experiencing supply tightness due to strong demand and expect additional volume to be available as Q3 progresses,” said Daniel Snyder, a spokesman for Intel.
[...]
It should be noted that nowadays the vast majority of personal computers are shipped by sea. It takes 42 days (including customs) for a ship to reach Europe from Taiwan, according to Asustek. Freight shipping time from Taiwan to the U.S. is around 20 days. As a result, after Asus (or other PC makers) gets the new processors and manufactures its computers, it will take at least 1.5 months for actual products to reach the market.

If Intel increases supply of “Skylake” products in late September, actual chips will show up in retail one to four weeks after it ships chips to its clients (i.e., in October). However, actual notebooks powered by “Skylake” processors will only get to the market in November.
The Intel spokesman says the shortage is because they are "experiencing supply tightness due to strong demand". But Intel usually has the sales channels filled at launch and is ready to handle strong demand. So I don't think that tells the complete story. Maybe they are still having troubles and yield issues with their 14 nm process as the article speculates on?

Finally I'm also a bit sad to see that we cannot expect Skylake notebooks on the market until November. I was planning to get one earlier than that. Oh well...
 
Last edited:

Sweepr

Diamond Member
May 12, 2006
5,148
1,142
131
myocardia said:
Here's his first attempt: http://www.eurogamer.net/articles/di...
5-6600k-review That's some shill using a 1.2 Ghz overclocked 6600k, along with a +25% overclocked memory controller, compared to a bone-stock i7-3770k and 4790k. The shill never even mentions what speed RAM he used with the IVB and HW, hence my "or even 1,333 Mhz DDR3" comment.

Another one of your lies. There's both stock and overclocked results for Core i5 6600K. On top of that, there's a clock per clock comparison at their Core i7 6700K review.
Eurogamer reviewed Devil's Canyon with a Corsair 1866/C9 kit and they most likely used the same kit here, another myth busted unless you have evidence to prove otherwise (which you don't, that's why you are attacking them by calling them 'shills' and 'payed by Intel').

Really? You mean that two different reviews, performed by different people, using different hardware, at difference times, could have a few percentage points of performance difference? Imagine that!

So now you admit there could be different results? What a surprise, you just called reviewers who dedicate hours to provide us the best data possible 'shills and payed by Intel' in the last pages (because they obtained better results from Skylake).

Here are the calculations I used from the above paragraph:
DDDR3 1,866 CL9= 9.64ns of latency
DDR4 2,133 CL 15= 14.06 ns of latency
DDR4 2,666 CL 15= 11.25 ns of latency
DDR4 2,666 CL 16= 12.00 ns of latency

No, you keep repeating this crap pretending it changes the facts.

DDR3-1600 CL8/CL9 DDR3 (used in most Haswell tests) is superior to DDR4-2133 CL15 DDR4 in true latency (used in AnandTech's Skylake review).

Haswell DDR3 1600 CL9 = 1.25 ns x 9 CL = 11.25 ns true latency
Skylake
DDR4 2133 CL15 = 0.94 ns x 15 CL = 14.06 ns true latency
DDR4-2400 CL15 = 12.5 ns latency
DDR4-2800 CL15 = 10.7 ns latency
DDR4-3000 CL15 = 10 ns latency

mikk just made you look like a fool above, proving latency matters too. If any of your desillusions were real Crucial wouldn't make a big deal about real latency calculation neither AnandTech would normalize clocks/CL latency in their methodology.

Are you questioning AnandTech's methods now? Funny cause you just posted benchmarks from their Skylake review to try to convince us latency wouldn't influentiate results (which you failed).

Even the most pedantic person I've ever met would conclude that a maximum of 3.1% difference, when comparing like speeds of RAM from different architectures (2,133 DDR4 vs 1,866 DDR3, 2,800 DDR4 vs 2,133 DDR3, DD4 3,600 vs DDR3 2,666, et al) and a minimum of .3% difference, are in fact in the zero percent to extremely close to zero percent range.

LOL, is this coming from the same person trying hard to convince us that reviews with minor RAM differences (when you normalize clocks/latency) are invalid and made by shills just because Skylake did pretty well in some of them? Double standards and goalpost moving at its best.

AnandTech said:
Normally in our DRAM reviews I refer to the performance index, which has a similar effect in gauging general performance:

DDR3-1600 C11: 1600/11 = 145.5
DDR4-2133 C15: 2133/15 = 142.2

As you have faster memory, you get a bigger number, and if you reduce the CL, we get a bigger number also. Thus for comparing memory kits, if the difference > 10, then the kit with the biggest performance index tends to win out, though for similar kits the one with the highest frequency is preferred.

PCLab's review
DDR4-2666/16 = 166 (Skylake-S)
DDR3-1866/9 = 207 (Haswell)

According to them, PCLab still favours Haswell.

Some Haswell tests:

DDR3-1866 CL8 beating DDR3-3000 CL12
55b%20CFX%20Sleeping%20Dogs%20Minimum_575px.png


DDR3-1866 CL8 beating DDR3-2933 CL12
54b%20CFX%20Tomb%20Raider%20Minimum_575px.png


DDR3-1600 CL8 beating DDR3-3000 CL12
24b%206950%20Dirt3%20Minimum_575px.png


...But but but you said 25% higher bandwidth would make a huge difference and latency doesn't matter.

Note how close DDR3-1866 CL9 and DDR3-2666 CL12 are.
Difference is, PCLab and Eurogamer used DDR4-2666 with much worse latency (CL16).
Claiming those kits favour Skylake in any significant way is a joke at best (maybe the opposite is true).

Skylake is 17% faster than Haswell per clock @ GTA V and Far Cry 4, throw in any memory DDR3 kit you want, Haswell would still lose. ;)
Stop thread crapping and get over it, Skylake is better than your Core i7 4790K.
 
Last edited:

witeken

Diamond Member
Dec 25, 2013
3,899
193
106
http://www.kitguru.net/components/c...al-shortage-of-intel-skylake-microprocessors/

The Intel spokesman says the shortage is because they are "experiencing supply tightness due to strong demand". But Intel usually has the sales channels filled at launch and is ready to handle strong demand. So I don't think that tells the complete story. Maybe they are still having troubles and yield issues with their 14 nm process as the article speculates on?

Finally I'm also a bit sad to see that we cannot expect Skylake notebooks on the market until November. I was planning to get one earlier than that. Oh well...
There will be Skylake before November. There will be Skylake in October and possibly September.

The reason isn't yield, but inventory.
 

Fjodor2001

Diamond Member
Feb 6, 2010
3,705
182
106
There will be Skylake before November. There will be Skylake in October and possibly September.
You mean notebook Skylake before November, and in actual products available in the stores? Or just more desktop inventory by then?
The reason isn't yield, but inventory.

Care to clarify that? You mean that Intel has lots of inventory of old Haswell / Broadwell CPUs that they want to clear out first? Or that the sales channels have lots of old inventory? Or both? :\

Also, from where have you heard this BTW?
 
Last edited:

Tovarisc

Member
Jun 12, 2015
50
0
0
Ditched that Intel software and redid OC.

i5 6600K @ 4.4GHz with core Volts being 1.3, after doing several hours of gaming + running RealBench system does seem stable.
 

CakeMonster

Golden Member
Nov 22, 2012
1,384
482
136
I think the Noctua mounting bracket is the most solid and the nh-d15 is also one of the best performing. It is very large though so you have to be careful of memory clearance etc. any reason you are ruling out an AIO water kit like Corsair H110?

I admit I haven't looked at water cooling in years but it always seemed to be issues with power consumption and noise back then. Is there a solution that's really simple to install, and doesn't require more than one fan, which can be turned down (like I do on air, way down), and with no pump noise that still cool better than air?
 

guskline

Diamond Member
Apr 17, 2006
5,338
476
126
Dave2150 thanks for the post. I just built a 4790k rig below with the same asus motherboard (Z97) at the same clock speed with DDR3-2400 ram.

Here is my Cinebench 11.5 result:
8x6g4p.jpg


Obviously Skylake has improved. Would be fun to compare more software benchmarks with you since you have the same clocked but skylake chip
 
Mar 10, 2006
11,715
2,012
126
http://www.kitguru.net/components/c...al-shortage-of-intel-skylake-microprocessors/

The Intel spokesman says the shortage is because they are "experiencing supply tightness due to strong demand". But Intel usually has the sales channels filled at launch and is ready to handle strong demand. So I don't think that tells the complete story. Maybe they are still having troubles and yield issues with their 14 nm process as the article speculates on?

Finally I'm also a bit sad to see that we cannot expect Skylake notebooks on the market until November. I was planning to get one earlier than that. Oh well...

Yeah, funny how the channel is filled with Z170 boards even with the "strong demand" but the actual Skylake chips are MIA.

Also, agree with you there on Skylake in notebooks. This really stinks.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
Still plenty of Skylake chips in europe. Specific boards are harder to find.

Next month should be the release of the NUC6I3SYK and NUC6I3SYH with i3 6100U. And Q4 for the NUC6I5SYK and NUC6I5SYH. It will be interesting to see if the i5 versions will be GT3e based or not.
 

myocardia

Diamond Member
Jun 21, 2003
9,291
30
91
Also getting a Skylake NUC as HTPC.
I have no idea whether you're one of the very few people who post here that uses their HTPC for playback only, but if you aren't, you'll want to wait for the Kabylake NUC. Skylake only has hardware HEVC decode. Kabylake is gonna have both HEVC encode and decode.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
I have no idea whether you're one of the very few people who post here that uses their HTPC for playback only, but if you aren't, you'll want to wait for the Kabylake NUC. Skylake only has hardware HEVC decode. Kabylake is gonna have both HEVC encode and decode.

Videos, NAS, music, server functions, jumphost.

The current i5 661 is way overdue for replacement. Both due to codec and power consumption.

Unless the i5 versions offers something exceptional. I will go the i3 route. The 6100U should be faster than the i5 661 and at 15W instead of 87W.
 

mrmt

Diamond Member
Aug 18, 2012
3,974
0
76
Yeah, funny how the channel is filled with Z170 boards even with the "strong demand" but the actual Skylake chips are MIA.

Also, agree with you there on Skylake in notebooks. This really stinks.

While the supply chain for chipsets and MBs is different and also has different lead times, I also think the likely culprit is Intel 14nm. Intel really delivered in terms of costs, I really doubt that either Samsung or TSMC could deliver Skylake at 4Ghz with 120mm^2. OTOH it seems that Intel pushed itself a little too hard in terms of how hard they pressed the current technology.

It's been almost a year after the first 14nm process and only today we are starting to see 14nm chips to appear in meaningful quantities, and even now we are hearing about issues in the supply chain. I can't recall the last time a 1-year old node at Intel got availability issues, and given that 10nm was also pushed back, it rather more probable that something more fundamental at Intel R&D pipeline is causing the problems than this snafu being a one off issue.
 

DigitalWolf

Member
Feb 3, 2001
108
0
0
Yeah, funny how the channel is filled with Z170 boards even with the "strong demand" but the actual Skylake chips are MIA.

Also, agree with you there on Skylake in notebooks. This really stinks.

What is "funny" is how all these retail 6700k's managed to show up in Europe. People who live there are the ones that have them or say they see them... and you can find listings for them which are almost all from Europe (aka on amazon market or ebay etc). Some don't really have much of a mark up and some do of course. Out of all the listings I've seen the only non EU one.. was from South Korea.

How long it takes a cpu to get to Asus put into a prebuilt package and then shipped to the US isn't really a factor on the inventory I'm looking for.

Some of the "builders" have 6700k's in the US because there are prebuilt systems "in stock" on newegg... or available for same day ship on some websites (ie: Cyberpowerpc etc)

Yet the version of the 6700k I'm looking for... is only really available in Europe.. its like all the Intel manufacturing is over there... or something. I'm glad someone got retail 6700k's I'm just not sure how North America was entirely skipped.. and then told there was a "shortage"... we never had product here other than for some OEM's.
 

richierich1212

Platinum Member
Jul 5, 2002
2,741
360
126
Yet the version of the 6700k I'm looking for... is only really available in Europe.. its like all the Intel manufacturing is over there... or something. I'm glad someone got retail 6700k's I'm just not sure how North America was entirely skipped.. and then told there was a "shortage"... we never had product here other than for some OEM's.

5775c and 5675C still unavailable as well.
 

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
It's been almost a year after the first 14nm process and only today we are starting to see 14nm chips to appear in meaningful quantities, and even now we are hearing about issues in the supply chain. I can't recall the last time a 1-year old node at Intel got availability issues, and given that 10nm was also pushed back, it rather more probable that something more fundamental at Intel R&D pipeline is causing the problems than this snafu being a one off issue.

Intel's had Broadwell-U and Y available in decent supply for awhile now. Of course probably only the 2+2 models which are 80-something mm2.

Broadwell-H is basically not available either.

What is "funny" is how all these retail 6700k's managed to show up in Europe

I'm don't how they would ship goods from Malaysia into Europe, but maybe it'd be possible to ship mostly on land which would be easier?
 

tential

Diamond Member
May 13, 2008
7,355
642
121
It is obvious from his post history isnt it?

It's painfully obvious why demand increases. As the FX line becomes less and less viable of an option, the ONLY thing being recommended on gaming forums are K line processors. It's no wonder it's hard to stock them. Just head over to reddit and see how many people are recommending intel.

In 2017, when Zen is out, I expect it to be SLIGHTLY harder for intel to sell their products withuot trying. I hope Zen is a viable alternative (or at least good enough that the fanboys buy and keep AMD alive), but when skylake sits at the top by itself... it's pointless to bring up a 3W TDP increase or whatever. No one cares. We care that it's the best chip out that we can get.
I'm excited for Skylake-E. I probably will purchase it no matter what and hand my spare parts over to my bro.
 

cmdrdredd

Lifer
Dec 12, 2001
27,052
357
126
I admit I haven't looked at water cooling in years but it always seemed to be issues with power consumption and noise back then. Is there a solution that's really simple to install, and doesn't require more than one fan, which can be turned down (like I do on air, way down), and with no pump noise that still cool better than air?

Well the Corsair H110i is supposed to be pretty good, but the reason I bring it up is because it allows you much more room around the memory slots and PCIe slots. Big tower coolers extend over them sometimes.