Kaby Lake - 5.0 GHz at 1.35 Volts reported.

[FlyIce 354]

I would say definitely not for 6700K. I consider from my 2500K to 7700K is a decent upgrade, but still from 4.9 GHz 2500K to 5 GHz 7700K I gained only 100MHz after nearly six years. This is pretty ridiculous.

 

I'm longing for the days that Intel or AMD will make 10GHz chips so all of us can double the fps effortlessly...  

[John_Tavendale 453]

Temps don't kill.....voltage kills.

This is true. It's kicking along nicely at 1.35v at the moment.

[martin-w 3,260]

 

I have a weird observation. Below is an XTU plot of a flight from Fly Tampa Sydney on the PMDG T7 from the gate to cruise. At 5 Ghz, AVX at -5 and 1.37 Vcore. The XTU maximum temperature is 89C while most of the time the temperatures are below 75C. Using CPUID on the same time, the max temp reported is 96C and using Real Temp, it is 90C. Which to believe?

 

 

 

 

 

 

As Hasse said... there's a difference between core temp and package temp. The temps come from different sensors. The temp of each core comes from a temp sensor in each core. The Package temp comes from a sensor embedded on the CPU package itself.  

Temps don't kill.....voltage kills.

 

 

 

But temps do cause throttling and eventual shutdown. And if it wasn't for the safety features of modern CPU's excessive temp would indeed damage chips. That's why they're designed to throttle and shut down.

sounds like these chips arent worth the upgrade coming from 6600/6700k. Im really looking forward to the amd ryzen and see what it can do. 

 

 

 

I don't think it was ever going to be. It's a refinement of Skylake. It's not a new architecture. It's just been fine tuned a bit and a few new features included.

 

Intel's "tick-tock" has been replaced with what they call Process Architecture, Optimisation. Basically a two stage development model has been replaced with a three stage development model. Two reasons I would say. 1. Performance improvements are getting harder to achieve. 2. Lack of competition from AMD.

[PaulGR 168]

Temps don't kill.....voltage kills.

 

How is that temps don't kill considering it is the temps that you worry about when you increase the voltage??

:blink:

[sizziano 172]

How is that temps don't kill considering it is the temps that you worry about when you increase the voltage??

:blink:

 

I think that was badly worded. Temps can of course kill a CPU but not like too much voltage. If the CPU hit's it's max temp it will throttle, sustaining high temps will eventually degrade a chip and possibly kill it but that is not what too much voltage will do. Over voltage will kill a CPU outright without much warning. So basically; high temps= slow degradation that will kill over time. Too much voltage= basically insta-death.

[martin-w 3,260]

 So basically; high temps= slow degradation that will kill over time. Too much voltage= basically insta-death.

 

 

 

 

CPU degradation is caused by electromigration, and electromigration has a relationship to temperature. Namely, lower CPU temp results in reduced electromigration and increased CPU temp results in increased electromigration. It's the metal interconnects that are impacted by electromigration. So basically, if the CPU is cold enough, higher voltage is less of an issue.

 

In regard to your hypothesis that excessive voltage results in instant CPU death: 

 

If we take Skylake as an example. Intel recommend, for daily use, no higher than 1.45 volts. The absolute max voltage on the other hand is 1.52 volts. So the question to ask is if we apply, lets say, 1.54 volts, or 1.58 volts, what will happen?

 

Two possibilities:

 

1. Heat increases as a result of the excessive voltage, the CPU's thermal protection can't react quick enough and the CPU is killed.

 

2. Heat increases as a result of the excessive voltage, the CPU thermal protection cuts in and the CPU shuts down safely. 

 

I'm not sure which of the above would apply. It would depend on the capabilities of the thermal monitor how fast it would react. I can't find anything on line that provides this information.  

[westman 257]

CPU degradation is caused by electromigration, and electromigration has a relationship to temperature. Namely, lower CPU temp results in reduced electromigration and increased CPU temp results in increased electromigration. It's the metal interconnects that are impacted by electromigration. So basically, if the CPU is cold enough, higher voltage is less of an issue.

 

In regard to your hypothesis that excessive voltage results in instant CPU death:

 

If we take Skylake as an example. Intel recommend, for daily use, no higher than 1.45 volts. The absolute max voltage on the other hand is 1.52 volts. So the question to ask is if we apply, lets say, 1.54 volts, or 1.58 volts, what will happen?

 

Two possibilities:

 

1. Heat increases as a result of the excessive voltage, the CPU's thermal protection can't react quick enough and the CPU is killed.

 

2. Heat increases as a result of the excessive voltage, the CPU thermal protection cuts in and the CPU shuts down safely.

 

I'm not sure which of the above would apply. It would depend on the capabilities of the thermal monitor how fast it would react. I can't find anything on line that provides this information.

Thanks Martin

 

If we set +1.6v we get a warning in bios 1.58v it boot normal, have run the Skylake at 1.56v short benches with custom water and 1.9v cooled with LN2

If you have Subzero cooling you boot in with 1.58v , when booted in to Win open Turbo V-Core do the settings,

If you cool the cpu with LN2 let the cpu cool down to approx -165C set the vcore to 1.85 to 1.9v and start bench.

XTU for example you get positive temp if you run 1.85v 6.7ghz you get a delta 190C.

 

If you have a killawatt or what the devices name in eng is, you can observate how the powerdraw decrease when cooling down the Cpu

 

Test one 7700K on a new Win7 pro install, only screen with high vcore, cooling custom water

https://dl.dropboxusercontent.com/u/54475361/OBT/7700k%20vcore.png

[laserit 93]

Well I must say that I'm delighted with my new 7700k. Hit F11 for the OC wizard in my bios, then turned of HT and i'm running 4.94 averaging 70c in Real Bench with the highest temp hitting 78c

 

P3d and XP are tuning in really nice. Going from a 4770k @ 4.6 to a 7700k @4.94 is very noticeable.   

[FlyIce 354]

quick question, for P3D, XP11 etc, should I turn off hyperthreading or to turn it off?

[martin-w 3,260]

Thanks Martin

 

If we set +1.6v we get a warning in bios 1.58v it boot normal, have run the Skylake at 1.56v short benches with custom water and 1.9v cooled with LN2

If you have Subzero cooling you boot in with 1.58v , when booted in to Win open Turbo V-Core do the settings,

If you cool the cpu with LN2 let the cpu cool down to approx -165C set the vcore to 1.85 to 1.9v and start bench.

XTU for example you get positive temp if you run 1.85v 6.7ghz you get a delta 190C.

 

If you have a killawatt or what the devices name in eng is, you can observate how the powerdraw decrease when cooling down the Cpu

 

Test one 7700K on a new Win7 pro install, only screen with high vcore, cooling custom water

https://dl.dropboxusercontent.com/u/54475361/OBT/7700k%20vcore.png

 

 

So the BIOS wont allow you to set a huge voltage? 

 

So what happens if you boot into windows at 1.58 or lower, and then in Turbo V-Core set a stupidly high voltage?

 

Will the CPU's thermal protection cut in, in time, and save the CPU from damage?

Or will the CPU's thermal protection not be quick enough to react and the CPU killed?

 

Asus broke the world record apparently, as I'm sure you've seen. 7.3 GHz! Voltage is very low though which surprised me, 1.205.

 

http://www.shacknews.com/article/98344/asus-breaks-overclock-world-record-on-intel-kaby-lake-cpu

Well I must say that I'm delighted with my new 7700k. Hit F11 for the OC wizard in my bios, then turned of HT and i'm running 4.94 averaging 70c in Real Bench with the highest temp hitting 78c

 

P3d and XP are tuning in really nice. Going from a 4770k @ 4.6 to a 7700k @4.94 is very noticeable.   

 

 

4.94 almost 5GHz at 78C in RealBench is very good Floyd.   :smile: Mind you, you are saving 10 degrees with HT off.

 

Which cooler are you using? And what voltage?

[westman 257]

So the BIOS wont allow you to set a huge voltage? 

 

So what happens if you boot into windows at 1.58 or lower, and then in Turbo V-Core set a stupidly high voltage?

 

Will the CPU's thermal protection cut in, in time, and save the CPU from damage?

Or will the CPU's thermal protection not be quick enough to react and the CPU killed?

 

Asus broke the world record apparently, as I'm sure you've seen. 7.3 GHz! Voltage is very low though which surprised me, 1.205.

 

http://www.shacknews.com/article/98344/asus-breaks-overclock-world-record-on-intel-kaby-lake-cpu

 

 

 

4.94 almost 5GHz at 78C in RealBench is very good Floyd.   :smile: Mind you, you are saving 10 degrees with HT off.

 

Which cooler are you using? And what voltage?

The WR was the vcore +1.95v , they have CPUZ set enable XOC (very light ) read the vid .

Normaly if you put in +1.95v and have not cold down the protektion kicks in.

Kill a CPU is very difficult , XTU is worst normaly the cores is ok but Its the memcontroller that die.

[Christopher Low 5,393]

Get back to me when Intel find a way to reach 7.3Ghz with stock air cooler :smile:

[OneWhoKnocks53 51]

I'd just like to share my recent experience with one particular 6700k.  The stock TIM between the die and the heat spreader is worse than I ever imagined!  I decided to delid and change the TIM out with Coolaboratory Liquid Pro, not Ultra.  I used Gelid paste between the spreader and water block of my H115i, same before and after.  Temps before the delid at 4.8GHz @ 1.32vCore - 71C average per core, 75C CPU package.  After the delid - 4.8GHz @ 1.32vCore - 54C average per core, 56C CPU package.  No joke, same stress test, same 6 hour run, same ambient temp.  Anyone here delidded a 7700k yet? I would love to know if the TIM used on Kaby is the same as Skylake.

[laserit 93]

4.94 almost 5GHz at 78C in RealBench is very good Floyd.   :smile: Mind you, you are saving 10 degrees with HT off.

 

Which cooler are you using? And what voltage?

 

I've now got her @5.15GHz with 1.36v 

 

top core temp = 76deg

top package temp =78deg

 

no HT

 

Cooling with a Corsair 100i v2

[martin-w 3,260]

The WR was the vcore +1.95v , they have CPUZ set enable XOC (very light ) read the vid .

Normaly if you put in +1.95v and have not cold down the protektion kicks in.

Kill a CPU is very difficult , XTU is worst normaly the cores is ok but Its the memcontroller that die.

 

 

Thanks Hasse.

I'd just like to share my recent experience with one particular 6700k.  The stock TIM between the die and the heat spreader is worse than I ever imagined!  I decided to delid and change the TIM out with Coolaboratory Liquid Pro, not Ultra.  I used Gelid paste between the spreader and water block of my H115i, same before and after.  Temps before the delid at 4.8GHz @ 1.32vCore - 71C average per core, 75C CPU package.  After the delid - 4.8GHz @ 1.32vCore - 54C average per core, 56C CPU package.  No joke, same stress test, same 6 hour run, same ambient temp.  Anyone here delidded a 7700k yet? I would love to know if the TIM used on Kaby is the same as Skylake.

 

 

Yep, that's not an unusual drop in temp for Skylake after delidding. Some have dropped as much as 30C.

 

And yes, Kaby Lake is no different. Report not long ago of a 30C drop in temp after delidding.

I've now got her @5.15GHz with 1.36v 

 

top core temp = 76deg

top package temp =78deg

 

no HT

 

Cooling with a Corsair 100i v2

 

 

 

Great result! Lucky boy. And to be honest there's no much that makes use of HT.