[OCN] Kabylake-S doing 5GHz on Air

[Arachnotronic]

From poster Oj010 on OCN on Kaby Lake-S:

Frequency fans rejoice, as these are doing 5 GHz on air and more - not just one or two, but ALL samples that have been tested. I think a terrible chip will still get around 4.8 GHz or so.

http://www.overclock.net/t/1610124/...kaby-lake-performance-teased/90#post_25482464

Hope this is legit, it's about time the 5GHz barrier for a high perf/clock chip be broken

 

[ozzy702]

From poster Oj010 on OCN on Kaby Lake-S:



http://www.overclock.net/t/1610124/...kaby-lake-performance-teased/90#post_25482464

Hope this is legit, it's about time the 5GHz barrier for a high perf/clock chip be broken

Does that particular poster have credibility?

 

[Arachnotronic]

Does that particular poster have credibility?

Not 100% hit rate, but he's definitely gotten some things right that appear to be more than just dumb luck.

 

[NTMBK]

Wow, nice if that turns out to be representative. 5GHz with Skylake IPC? That's going to kick some serious butt.

 

[Azuma Hazuki]

Niiiiiiiice! I'm still probably going to be getting an 8C/16T Zen just because of what my needs are, but this is awesome for the mainstream and for serious gamers. Intel's really perfected its 14nm process

 

[Arachnotronic]

Honestly kind of regretting my move back to HEDT. Shoulda just kept my z170 setup so I could upgrade to this. Dang.

 

[strategyfreak]

http://wccftech.com/intel-core-i7-6700k-skylake-overclocked-52-ghz-air-cooling-135v/

How did that turn out? A 6700k at 5.2ghz on air would be an extremely rare find today.

I wouldn't buy any of these reports or at least take them with a serious grain of salt. Intel could be sending out selected ES chips known to clock very high to impress people. This guy could just be flat out lying. The only way to find out for sure how the chips will overclock is to wait until the public gets access on a large scale. I don't buy prerelease rumors from neither AMD nor Intel because they're mostly released to generate hype and attention, not show their products in an objective fashion. The best strategy is to wait until its released.

Honestly kind of regretting my move back to HEDT. Shoulda just kept my z170 setup so I could upgrade to this. Dang.

lol

 

[PPB]

That is the same "Polaris cant go beyond 850mhz" guy. Tons of salt need to be applied before taking anything he says at face value.

 

[scannall]

Like with all the Zen stuff, just wait till it's actually out and tested. Not that speculation isn't fun from time to time.

 

[Arachnotronic]

Like with all the Zen stuff, just wait till it's actually out and tested. Not that speculation isn't fun from time to time.

True, but Zen is a relatively unknown quantity. We know Kabylake is Skylake on a new process, so it makes sense that it would be able to clock better than Skylake. Whether it can hit 5GHz+ regularly remains to be seen, but it's at least it's something that's reasonable to hope for.

 

[Ajay]

Isn't Kabylake using Intel's updated 14nm plus process? If so, then perhaps the numbers posted by this guy aren't that far off. 14+ sounds like a lower density, higher clocking process from the little I've read about it.

 

[CHADBOGA]

I'm still very happy running my Ivy Bridge i5 3570K @ 4.0Ghz, but it is good to see the 5.0Ghz barrier looking like it is going to be breached by Intel for their mainstream quads.

Whenever I do my next CPU upgrade, I think I would like to have 6 cores running at 6Ghz, with only a moderate increase in voltage, so I still think 2020 or beyond for me.

 

[Arachnotronic]

I'm still very happy running my Ivy Bridge i5 3570K @ 4.0Ghz, but it is good to see the 5.0Ghz barrier looking like it is going to be breached by Intel for their mainstream quads.

Whenever I do my next CPU upgrade, I think I would like to have 6 cores running at 6Ghz, with only a moderate increase in voltage, so I still think 2020 or beyond for me.

6GHz is only a 20% increase from 5GHz, and yet something tells me it's not going to be happening by 2020...or pretty much ever

 

[Roland00Address]

http://wccftech.com/intel-core-i7-6700k-skylake-overclocked-52-ghz-air-cooling-135v/

How did that turn out? A 6700k at 5.2ghz on air would be an extremely rare find today.

I am merely guessing. (Note this part is for sure known) But kabylake and skylake are not the same dies even though effectively they are the same architecture but kabylake also has a better gpu.

One of the things they did is tweak the process but they also modified minor parts of the chip not changing how the chip works but instead altering minor parts of its layout. One of the things they mentioned for mobile is that they increased the height of the finfets.

http://www.anandtech.com/print/1061...s-six-notebook-skus-desktop-coming-in-january

(Note this part is for sure known) This in turn allows intel to do 500 mhz higher turbos for all the core m3 /i5/i7 / 4.5 w tdp Y series. This part is a guess I would not be surprised if a similar mhz burst maybe 500 mhz, maybe 400, maybe 200, maybe 600 also occurs on desktop due to these tweaked process and tweaked design. I would not be surprised this redesigned process and processor design with the same effective architecture but material tweaks will allow higher clock speeds at the 4ghz + range.

For example taller fins means you can run a chip at a higher clock speed with a lower voltage than the default voltage (with smaller fins) for the "electrons" move from one end of the chip to the other and completely clear out faster and there is less leakage due to clockspeed at a specific voltage. (I believed I said that correctly)

 

[moonbogg]

I thought kabylake was going to use the new HEDT platform. It is right? Isn't that the chip that's hitting 5ghz? Also, the conundrum of fast quad vs slower hex+ still sucks. Wish we could have the best of both worlds.

 

[frozentundra123456]

Yea, I hope we do see fairly common 5ghz overclocks, but I remain skeptical as well. (Unpleasant memories of how Devil's Canyon was claimed to hit 5ghz on air). Even Sandy Bridge, fondly remembered as the best overclocker of all intel chips, I dont think was easy to get 5ghz unless you got a better than average chip.

 

[CHADBOGA]

6GHz is only a 20% increase from 5GHz, and yet something tells me it's not going to be happening by 2020...or pretty much ever

Intel's 14+nm process has made me optimistic.

As I still hope to get 10 years out of my i5 3570K, I think that would mean a 2022 purchase.

 

[VirtualLarry]

As I still hope to get 10 years out of my i5 3570K, I think that would mean a 2022 purchase.

That's still a GREAT CPU (and a fairly conservative OC).

I might get a 7700K, if they do (commonly, not extremely cherry-picked ES) OC to 5.0Ghz.

I remember when I got my Athlon XP, and joined the "1Ghz+ club". It was an awesome feeling.

Edit: I still want to get an AM4 Zen 8C/16T throughput monster for DC work, rather than gaming, too.

 

[aigomorla]

Does that particular poster have credibility?

from the way that thread sounds no.

it could also be that he is under so much NDA's that he cant say anything at all however i think he is a intel fanboy just spewing out garbage.

You can tell from his avatar being a ES tag. lol..

 

[Arachnotronic]

That is the same "Polaris cant go beyond 850mhz" guy. Tons of salt need to be applied before taking anything he says at face value.

At the very least, he seems to have owned up to his error:

I haven't been around much for the last three weeks as I started a new job and haven't had much time for forums, but I'm online now.

So, turns out my Polaris 10 in October story was wrong. Very wrong. I don't know how or why, I don't know if my source was fed incorrect information to finger himher out or if *I* was fed incorrect information, or even if the information was correct but incomplete (I've heard that AMD has potentially changed fab for Polaris 10 away from GloFo - haven't had time to even look into it so no comment there, it could also be that it's Vega coming in October and the story had nothing to do with Polaris 10 at all). Heck, with current reports of the architecture topping out at 1,400 MHz maybe AMD swapped fabs and rushed the launch, who knows. I haven't had time to look into it at all and won't for a while still, it'll be a few more weeks before I'm settled in comfortably at work.

This was the first time I've leaked a story from said source, and I won't be using him/her again regardless of previous correct information as this was a pretty major thing to get wrong.

Point is, although I firmly believed the information given as that particular source has been correct several times before, it was me that pushed the story hard and not him/her. Over here (http://www.overclock.net/t/1599950/...-might-launch-in-october-now/90#post_25154222) I agreed to an infraction if it turned out that the October launch was incorrect and I haven't forgotten. Mods do what you deem necessary.

Anyway, as far as this guy's track record, here are some recent predictions.

Before Broadwell-E launched (May 30 post), he predicted that it was an overclocking dud:

So Broadwell-E isn't clocking very well, 5.2-5.3 GHz on LN2 with a fairly weak IMC :/

On May 12, he made several posts predicting impressive frequencies for GTX 1080:

GTX 1080 can reach overclocks that Maxwell couldn't even dream of. The frequencies are going to make you laugh.

I have to bite my tongue so hard right now because I'm not allowed to say much, but 1.73 GHz is NOTHING.

He also threw cold water on the idea of 2500MHz GTX 1080s on less-than-exotic cooling:

I bolded two lines by accident. 2,500 MHz isn't going to happen on water without divine intervention. LN2 yes, and far beyond.

So, like I said, not a 100% hit rate, but certainly not talking out of his rear end, IMO.

 

[witeken]

6GHz is only a 20% increase from 5GHz, and yet something tells me it's not going to be happening by 2020...or pretty much ever

Maybe when they find the elusive room temperature superconductor .

Edit: Don't really like the smilies all that much.

 

[Ajay]

Shocking - stupid rumor is stupid. The internet if full of airball wizards
I guess we'll just have to wait an see.

New emojis - apparently we are meant to believe that less is more *shrug*.




edit: typo

 

[Doom2pro]

From poster Oj010 on OCN on Kaby Lake-S:



http://www.overclock.net/t/1610124/...kaby-lake-performance-teased/90#post_25482464

Hope this is legit, it's about time the 5GHz barrier for a high perf/clock chip be broken

I don't think this poster has credibility, sounds like an attention seeking Intel fan. Listen to Dresdenboy on these forums, he is the real deal.

 

[IntelUser2000]

Sandy Bridge 5Ghz: https://www.google.ca/search?q=sand...aLKV9T8Ic3N-APSs4mQDA#q=sandy+bridge+air+5ghz
4770K 5GHz @ 0.9V: http://www.tweaktown.com/news/30292...cked-to-5ghz-on-just-0-9v-of-power/index.html
4790K 5GHz: https://www.google.ca/search?q=sand...cr&ei=UaLKV9T8Ic3N-APSs4mQDA#q=4790k+air+5ghz
Broadwell C 5GHz: https://www.google.ca/#q=broadwell+air+5ghz
Skylake 5GHz: https://www.google.ca/#q=skylake+air+5ghz

With Devil's Canyon, Intel embraced the idea of flaunting that the chip can reach 5GHz on air. When it was found out that it couldn't do that reliably, they shut up. Likely they continued to viral marketing tactics to continue to spread such rumors with Broadwell and Skylake.

 

[SAAA]

Sandy Bridge 5Ghz: https://www.google.ca/search?q=sand...aLKV9T8Ic3N-APSs4mQDA#q=sandy+bridge+air+5ghz
4770K 5GHz @ 0.9V: http://www.tweaktown.com/news/30292...cked-to-5ghz-on-just-0-9v-of-power/index.html
4790K 5GHz: https://www.google.ca/search?q=sand...cr&ei=UaLKV9T8Ic3N-APSs4mQDA#q=4790k+air+5ghz
Broadwell C 5GHz: https://www.google.ca/#q=broadwell+air+5ghz
Skylake 5GHz: https://www.google.ca/#q=skylake+air+5ghz

With Devil's Canyon, Intel embraced the idea of flaunting that the chip can reach 5GHz on air. When it was found out that it couldn't do that reliably, they shut up. Likely they continued to viral marketing tactics to continue to spread such rumors with Broadwell and Skylake.

There have been quite a lot of rumors about 5GHz over the years and many were untrue indeed, but you have to admit that a lot of Sandy Bridges did reach that easily even on air.
It was 22nm and 14nm that overclocked less, now 14nm+ seems finally promising, to the point that a stock chip on air is already at 4.5GHz (7700K at 95W) and another one is coming with even higher speed (Kaby Lake X at 112W).
I doubt that if the latter reaches say 4.8GHz at stock it won't overclock just 200MHz or more, plus given current Skylake average overclocks, let's put it at 4.6GHZ, a less than 10% increase could make 5GHz common again. They stated the process allows for 12% more performance, that seems to hold true for both 3GHz and 4GHz parts (core-M as well as high end quad got a 300-400MHz increase) so the same increase moving toward 5GHz isn't that impossible.