Intel Skylake / Kaby Lake

[lolfail9001]

6700K also hit 6.8Ghz with all cores and HT active, on LN2.

Where?
HWBot Cinebench rankings lack even 6.7Ghz okay, found one, all cores + HT active, let alone 6.8 (still have not found it).

 

[mikk]

http://www.facebookhk.com/

Retail i5-7600k pictured.

 

[Sweepr]

http://www.facebookhk.com/

Retail i5-7600k pictured.

That's not the only interesting bit there. Core i5-7600K doing 5.1 GHz?

After a successful upgrade bios, boot sure. The first thing is to turn off all unnecessary things such as LAN, USB and audio additional wafer. The CPU multiplier and voltage to minimum, and then gradually upwards. Results obtained up Fengyun 7600K clocked at 5.2Ghz, but not very stable. So down to 5.1G, the required voltage is about 1.55v. Although the frequency is not high, the voltage is only general. But it's very good temperature, idle at less than 30 degrees Celsius! Bios which reads out the number as well.

 

[LTC8K6]

Where?
HWBot Cinebench rankings lack even 6.7Ghz okay, found one, all cores + HT active, let alone 6.8 (still have not found it).

http://www.kitguru.net/components/c...th-all-cores-active-on-msi-titanium-platform/

 

[coercitiv]

That's not the only interesting bit there. Core i5-7600K doing 5.1 GHz?

While I'm quite optimistic about Kaby Lake poking 5Ghz, these early "leaks" only make me remember the 5Ghz Broadwell ES overclocks. Seems to me all fans want to believe this crap, no matter the brand.

 

[crashtech]

Every time we hear it said, we'll see lots of 5GHz chips! Someday this will be true, right?

 

[lolfail9001]

http://www.kitguru.net/components/c...th-all-cores-active-on-msi-titanium-platform/

Never mind, you talk about verification only. You know my opinion of those

Also, 1.5V for 5.1Ghz AND under 30 degrees idle? Something is very wrong here.

 

[cbn]

Kaby Lake-S (desktop) engineer samples out in the wild!

- Pentium G4620 @ 3.8 GHz
- Core i3-7300 @ 4.0 GHz

(snip)

Pentium and Core i3 reaching the magic 4.0 GHz.

I am assuming the 4.0 Ghz Kabylake Core i3 7300 would have the same pricing as the 3.8 Ghz Skylake Core i3 6300 (ie, $138 to $147) and the 3.8 Ghz Kabylake Pentium G4620 (the first LGA 115x Pentium with hyperthreading) would have the same pricing as the 3.6 Ghz Skylake Pentium G4520 ($86 to $93).

But I wonder if there will also be 2C/4T Kabylake Pentiums offered at the 3.3 Ghz Skylake G4400 ($64) and 3.5 Ghz G4500 ($75 to $82) price points?

 

[VirtualLarry]

But I wonder if there will also be 2C/4T Kabylake Pentiums offered at the 3.3 Ghz Skylake G4400 ($64) and 3.5 G4500 ($75 to $82) price points?

That would be sweet. I would be interested in how viable those CPUs would be for gaming. Would they be just as good as the Skylake i3-6100, or would the lesser cache affect the performance notably?

 

[cbn]

That would be sweet. I would be interested in how viable those CPUs would be for gaming. Would they be just as good as the Skylake i3-6100, or would the lesser cache affect the performance notably?

Good question.

Remember the Kabylake Pentium 2C/4T doesn't have AVX, AVX2, TSX and FMA3 like the Core i3 6100 does.

So less instruction sets and 1MB less cache.

But you are right, if such a $75 to $82 Kabylake 2C/4T Pentium existed it should have the same 3.7 Ghz clock as Core i3 6100.

 

[VirtualLarry]

I just hope that they show up at the Pentium price-point, and not the i3 price-point.

2C/4T would be sweet for $60-70.

Then again, if the top-end consumer chips are going to be hex-cores as rumored, then Intel needs to start being a little more generous with cores and threads now, imho.

 

[witeken]

What I'm actually quite surprised by is that Intel just does not seem to respond in any way to the 10nm ramps that are going on in Q4 here of Samsung and soon TSMC.

They continue keeping node names that make themselves look uncompetitive and they don't seem aggressive at all about getting 10nm sooner to market to respond to TSMC and Samsung's ramps. Of course they don't have to, there's not much overlap and likely there won't be 10nm foundry products soon, but still .

 

[jpiniero]

Good question.

Remember the Kabylake Pentium 2C/4T doesn't have AVX, AVX2, TSX and FMA3 like the Core i3 6100 does.

So less instruction sets and 1MB less cache.

That's not true, the Pentiums typically do have the same L3 as the i3 x100 do (3 MB). It's the x300 that have 4 MB. The only difference would be AVX.

 

[LTC8K6]

Yes for the extra $20, going from 6100 to 6300 i3 you get a clock speed bump, a graphics clock speed bump, and 1MB more cache.
The Pentiums tend to just lack a little clock speed, some instructions, and HT, from the i3.

 

[Nothingness]

Remember the Kabylake Pentium 2C/4T doesn't have AVX, AVX2, TSX and FMA3 like the Core i3 6100 does.

Intel are really dumb and things are not improving it seems. I guess they'll wait for AVX to be 10 years old before supporting it everywhere. Their market segmentation stupidity has no limit

 

[Sweepr]

Core i5-7600K spotted!

http://www.expreview.com/50070.html

 

[Phynaz]

LN2........ so only 0.0001% will care.

WTB air/water results

FTFY

 

[coercitiv]

What I'm actually quite surprised by is that Intel just does not seem to respond in any way to the 10nm ramps that are going on in Q4 here of Samsung and soon TSMC.

Intel is just sitting up in the balcony and watching two dudes Kung fu fighting in the street, mocking each others skill. He may come down and teach them a lesson or two, but why bother with their lesser technique? Sure, a day may come when one of them may come up and challenge the old master, but for now all they can do is brag in front of the gullible crowd.

I suspect Intel is letting Samsung and TSMC have their paper nanometer war in front of the investors. Was it not you who were left unimpressed with the new claims Samsung made for their 10nm process? I confess I'm patiently waiting for a more detailed analysis from our more knowledgeable forum members.

 

[LarsSvenn01]

86% faster than Core i7 965.

I don't see the logic in that comparison when reading about the difference here. Really dosent sound like that big difference to me.

3DMark CPU Score
http://www.pcfrm.com/wp-content/uploads/2015/04/Intel-i7-6700K-3D-MARK-4.png

And i don't understand 3Dmark cpu score? If i remember correct i compared my GTX 680 2gb to a GTX 960 and the 680 Scored about 50% higher? Seems pretty equal in fps comparisons.


http://www.3dmark.com/3dm11/11668041
Also my i7 920 @ 3,8 ghz = 8373 physics score
First i5 6600k @ 3,5 ghz i find = 4 530
Second i5 6600k @ 3,5ghz = 4 379

Im having a hard time comparing my old cpu with the new ones, would hope to be a little more enlightened

I heard people saying there is a big difference, and i find it logical to be a big difference from 8 years newer technology. But i just cant find any clear reliable comparisons

 

[LTC8K6]

I don't see the logic in that comparison when reading about the difference here. Really dosent sound like that big difference to me.



And i don't understand 3Dmark cpu score? If i remember correct i compared my GTX 680 2gb to a GTX 960 and the 680 Scored about 50% higher? Seems pretty equal in fps comparisons.


http://www.3dmark.com/3dm11/11668041
Also my i7 920 @ 3,8 ghz = 8373 physics score
First i5 6600k @ 3,5 ghz i find = 4 530
Second i5 6600k @ 3,5ghz = 4 379

Im having a hard time comparing my old cpu with the new ones, would hope to be a little more enlightened

I heard people saying there is a big difference, and i find it logical to be a big difference from 8 years newer technology. But i just cant find any clear reliable comparisons

Well, compare 8 thread cpu to 8 thread cpu first. Obviously an 8 thread i7 CPU will have an advantage over a 4 thread i5 CPU in most things that are multi-threaded.

 

[LTC8K6]

Here you see an i3-6320 and an i7-920 swapping leads between single and multi threaded benchmarks.

When we stay with single threaded, the i3 has a big lead.
When we go multi, the i7 either catches up, or pulls in front of the i3.

http://www.anandtech.com/bench/product/47?vs=1783

If we change the modern i3 to a modern i7, the 920 is left in it's wake.

http://www.anandtech.com/bench/product/47?vs=1554

 

[witeken]

Damn, I just spent an hour on AnandTech's 10nm Samsung article correcting everyone.

I guess I take it too personally, but it just saddens me that even though there is so much talk about those nodes, almost NO one (except for me obviously^^) can get their facts straight.

And even when people do use sources, they are from incredibly poor sources like SemiWiki that spread misinformation.

 

[witeken]

Intel is currently holding Q3 earnings call.

 

[LarsSvenn01]

Alright thats an obvious thing i focused little on lol, made shit clearer there.

I also like those comparisons.

I would like to also compare
45nm 4c/8t (2008)
14nm 4c/8t (2016)
both for example @ 4ghz.

Copre 6c/12T @ 3,4ghz. Not really that impressed
http://www.anandtech.com/bench/product/1727?vs=444

Struggling understanding and comparing power in the cpu caused by the development the last decade other then lower TDP and shorter/faster signals.

 

[witeken]

BK at webcast:

3D XPoint qualified in Q4, thousands of samples to customers.
3D XPoint ramp in 2017. (He said 2018 but he made a specific note that he made a mistake, so the volume ramp is 2017.) Cost will come down through the year.

Purley sampling to customers, will integrate silicon photonics and Omni-path architecture. There will be a second generation with 3D XPoint.