6700Ks getting better?

[Arachnotronic]

So I recently picked 6700K + ASUS Maximus VIII Hero board, my second 6700K + Z170 board combo. My first 6700K came set at a stock voltage of 1.312V and was not a good overclocker -- I run it at 4.6GHz (and it took a lot of work to get it there).

The one I just set up comes out of the box with a voltage 1.248v. I was able to overclock it to 4.7GHz with minimal hassle, and as of writing it passed 24 hours of Intel Burn Test.

Although this may be due simply to the "silicon lottery" I'm inclined to think that as 14nm yields have improved (6700K is now readily available), Intel is able to put out chips that can meet the 6700K spec without having to crank up voltage as much.

 

[VirtualLarry]

That's good news. I wonder if newer CPUs (locked ones) will then be able to OC better too?

Debating on getting some i5-6500 CPU(s) in March.

 

[mysticjbyrd]

Why are yields an issue, when they have been on 14nm so long? Did they just underestimated the demand?

 

[ThatBuzzkiller]

Just silicon lottery ...

They didn't change their transistor specs or anything of that sort ...

 

[Dresdenboy]

So I recently picked 6700K + ASUS Maximus VIII Hero board, my second 6700K + Z170 board combo. My first 6700K came set at a stock voltage of 1.312V and was not a good overclocker -- I run it at 4.6GHz (and it took a lot of work to get it there).

The one I just set up comes out of the box with a voltage 1.248v. I was able to overclock it to 4.7GHz with minimal hassle, and as of writing it passed 24 hours of Intel Burn Test.

Is there a measurable performance difference (due to thermal limitations, etc.) between both chips?

 

[ShintaiDK]

My 6700K predates both yours I guess. Running at stock.

1.168V using the CPU-Z stresstest and CB11.5. 1.184V running Prime95.

Using BIOS 1.5 on my MSI Z170I Gaming Pro AC.

But mine got lower with a new BIOS. I can see with previous BIOS it was up at 1.240V.

 

[Arachnotronic]

Well then, I guess my original 6700K was a turd

 

[cytg111]

"Oh look another hit and run post from Arachnotronic showing how Intel is awesome!!" - cough cough.

Seriously thou'? How does that work out?

"I'm inclined to think that as 14nm yields have improved "

- With two data points? - You put in the coordinates, flip out the ruler, draw a straight line, A->B, flip out the old notes on linear regression, crunch the numbers. Yes. P is defn. less that 0.05! The math checks out! - And thus arriving at the mental state of being 'inclined to think' that Intels magnificent 14nm yields must have improved ? .. just wondering .


Trolling is not allowed here
Markfw900

 

[lehtv]

it passed 24 hours of Intel Burn Test.

Like... What? What's the point? :'( Poor CPU... why?

 

[Arachnotronic]

"Oh look another hit and run post from Arachnotronic showing how Intel is awesome!!" - cough cough.

Seriously thou'? How does that work out?

"I'm inclined to think that as 14nm yields have improved "

- With two data points? - You put in the coordinates, flip out the ruler, draw a straight line, A->B, flip out the old notes on linear regression, crunch the numbers. Yes. P is defn. less that 0.05! The math checks out! - And thus arriving at the mental state of being 'inclined to think' that Intels magnificent 14nm yields must have improved ? .. just wondering .

What the hell? First of all, unlike a certain poster I don't just stick a few copied and pasted charts with zero original commentary. I observed something and made a post detailing my observation.

Second, until recently 6700k chips have been hard to find, suggesting yield problems. Now they are available in any quantities anybody could reasonably want, suggesting that Intel is now better at making them.

I then noticed that a newer chip that I purchased is better than mine I purchased previously, which I thought may be as a result ofIntel's parametric yields getting better. Note that I also suggested in the OP that my conclusion might not be correct, and some posters provided evidence to suggest that this might be the case.

Oh and unlike a certain poster, I respond to the threads I start instead of starting a thread and running off.

 

[cytg111]

Dude, that was way too easy

 

[jpiniero]

Second, until recently 6700k chips have been hard to find, suggesting yield problems. Now they are available in any quantities anybody could reasonably want, suggesting that Intel is now better at making them.

Yeah, but the price (at least in the US) is still around $400-410, which unless Intel mandated the hike is still a rather large gouging for a product that was released nearly 4 months ago.

The 6700K as it stands is a really poor value compared to the 5820K or the 4790K+H97.

 

[Magic Carpet]

Do you know what sucks Beavis?

We can't buy pre-binned, low-voltage 6700K from Intel. I hate that lottery.

@ShintaiDK

Nice one :thumbsup:

On the positive note, I've just ordered another three Haswell E3 Xeons, maybe I'll keep one with the lowest voltage for myself. 6700Ks are still a bit expensive for my liking.

 

[Sweepr]

Better yields or not, nice to know you got a better chip.

What the hell? First of all, unlike a certain poster I don't just stick a few copied and pasted charts with zero original commentary. I observed something and made a post detailing my observation.

Pretty sure you're not a paid shill.

 

[ehume]

Interesting. I got an early i7 860. Then another 6 mo later. The first hardly overclocked -- I run it stock now. The other ran at 4GHz until I accidentally killed it. I bought an i7 4790k, then another a year later. The second overclocked 100MHz faster at a lower Vcore. Now you are reporting an improvement in your later-purchased CPU. I think in general we can expect improvements as Intel gets more time on any particular die.

 

[ShintaiDK]

Yeah, but the price (at least in the US) is still around $400-410, which unless Intel mandated the hike is still a rather large gouging for a product that was released nearly 4 months ago.

The 6700K as it stands is a really poor value compared to the 5820K or the 4790K+H97.

My favoured retailer have been 365-370 all the time.

Global prices should make it obvious its not a manufactor controlled price, but retailer.

 

[ShintaiDK]

Do you know what sucks Beavis?

We can't buy pre-binned, low-voltage 6700K from Intel. I hate that lottery.

@ShintaiDK

Nice one :thumbsup:

On the positive note, I've just ordered another three Haswell E3 Xeons, maybe I'll keep one with the lowest voltage for myself. 6700Ks are still a bit expensive for my liking.

Well it has to be said I have no idea how mine OC. It could be an OC turd for that matter :biggrin:

But the chip do fit my stock usage better than what Arachnotronic got. However, remember the changes through BIOSes as well.

 

[DrMrLordX]

Like... What? What's the point? :'( Poor CPU... why?

Nothing wrong with 24-hour stress testing. That's standard operating procedure for finalizing a stable overclock.

My 7700k has seen 24-hour runs of Prime95 Blend and AMD Linpack, and it's fine.

 

[Magic Carpet]

Well it has to be said I have no idea how mine OC. It could be an OC turd for that matter :biggrin:

Well your setup is certainly capable of that. Maybe when you get bored, might give it a spin

But the chip do fit my stock usage better than what Arachnotronic got. However, remember the changes through BIOSes as well.

Yeah, default given vCore varies from board to board, bios to bios. Your board seems to be quite power-efficient (and not just because it's MSI). When it happens, I usually keep it. It's rare (when every component seems to play nice with each other). I remember when I had two similar Thuban chips, and one had much lower vCore, I was even able to cool it with the stock heatsink reasonably quiet, while the other consumed more watts and required an aftermarket heatsink even for stock clocks for quiet operation.

You need volume, to get and to choose the best chip; or just get extremely lucky

 

[Arachnotronic]

Like... What? What's the point? :'( Poor CPU... why?

The CPU is just fine

Seriously though I can't accept any instability in any of my systems, so I have to go to extremes to make sure my overclocks are 100%.

 

[CakeMonster]

My 6700K from when most forum members were still complaining about availability overclocks great.

Not really sure why it would be interpreted as a "good thing" if yields have improved, since that implies that the CPU's weren't good to begin with. In my anecdotal experience the early ones OC very well.

 

[DrMrLordX]

The CPU is just fine

Seriously though I can't accept any instability in any of my systems, so I have to go to extremes to make sure my overclocks are 100%.

Darn tootin. Though I will admit, it's getting harder to guarantee stability out of overclocks, thanks to the increased complexity of what is included on that die. It's still doable.

Not really sure why it would be interpreted as a "good thing" if yields have improved, since that implies that the CPU's weren't good to begin with. In my anecdotal experience the early ones OC very well.

From the end user's perspective, we'll never know. Yields might have been awful, forcing Intel to release only the most "cherry" chips. Or maybe yields just weren't that bad thanks to the learning experience that is/was Broadwell. As time goes on, the 14nm process should become more mature, improving matters overall. I think that's the general sentiment being presented here, even if we do not have sufficient evidence to verify that improved process maturity is affecting overclocks of retail samples of the 6700k.

 

[kami]

Mine has a similar stock voltage (also a recent purchase). What voltage do you use to get 4.7? Mine is 1.325 adaptive with LLC 5 and it is stable so far (a stress test pushes it to 1.350-1.375, but normal load is 1.325, idle 0.8). D14 has no problems cooling it. Adaptive voltage was a disaster until I updated the bios though... it was trying to kill my chip with insane voltages.

 

[FlanK3r]

I tested about 10 chips of K Skylakes
1)one 4.6 GHz at CInebench at air as maximum 1.48V (impossible to cool it with higher voltage)
2) 5 pieces between 4.7-4.8 GHz 1.488-1.525V
3)3 pieces 4.9 GHz Cinebench 1.525V
4)one 5 GHz Cinebenchs at air 1.488V as maximum

 

[Arachnotronic]

I tested about 10 chips of K Skylakes
1)one 4.6 GHz at CInebench at air as maximum 1.48V (impossible to cool it with higher voltage)
2) 5 pieces between 4.7-4.8 GHz 1.488-1.525V
3)3 pieces 4.9 GHz Cinebench 1.525V
4)one 5 GHz Cinebenchs at air 1.488V as maximum

Wow, so my first Skylake really is a piece of crap

Thanks for sharing this info! Wish I had the cashola to buy a bunch of 6700Ks to test out, please post here once you get your hands on BDW-E!