FEATURED *OFFICIAL* Kaby Lake Overclocking results thread!!!

[Woomack]

I know that you know ... weird is only that 5.6 and 5.7 was about as hard to make, 5.8 regardless of voltage had problems to pass anything. Simple tests like hwbot prime are usually passing 100-200MHz higher than Cinebench but this time couldn't. The same clock for Cine 11.5 and something like SPi 1M or Pifast. XTU and Cine15 at 5.6GHz. My last 6700K required the same voltage for 5.2GHz.

 

[EarthDog]

We know... but I ask, as seasoned overclockers, why can't we manage to capture a screenshot with the right clocks in it. It's like a disease in some of our comp threads!

 

[Johan45]

We know... but I ask, as seasoned overclockers, why can't we manage to capture a screenshot with the right clocks in it. It's like a disease in some of our comp threads!

Just laziness

 

[trents]

Just got my 7600k and ASRock Z270 Killer SLI/ac a few days ago. I'm running it 24/7 at 4.8/1.3 vcore (bios setting) which seems to be the sweetspot but it will do 5.0 ghz stable.

 

[wingman99]

Looks like you have a good chip.

 

[trents]

Looks like you have a good chip.

Meh. The voltages you see there don't show the vdrop/vdroop. Had to set the vcore at 1.39 in bios and the LLC to highest level.

 

[Tgrable]

Im pretty bummed on my 7700k. I can get it to 5ghz at about 1.285 but at that point the temps are so freaking high I wont keep it there. If I even start moving my voltage to 1.2 my temps jump into the low 90's in prime95 and sadly it is 100% the Tim between my die and IHS.

I have checked several times to make sure I had good Tim coverage between IHS and waterblock. I pulled out my block and checked to make sure the internals were clean... made sure my water temps and flow was good... 3c delta T on full load! So my only conclusion is that the chip needs a delid to fix the terrible job intel did on making a good connection to die and IHS.

 

[trents]

Yes, it sounds like you have a reason to delid. They make a nice tool to push the lid off now. I delidded several Haswells without any issue using the "vise and whack" method. I've got a 360mm 30 fpi radiator with 4000 rpm 38mm thick Delta fans (controlled by a speed controller) and I had to turn the fans up pretty high to keep my temps down where you see at the voltage that allowed for 5 ghz. I was almost on the edge.

 

[wingman99]

Meh. The voltages you see there don't show the vdrop/vdroop. Had to set the vcore at 1.39 in bios and the LLC to highest level.

What I mean by a good chip is, I see people that can't get a stable clock at 5.0GHz.

 

[Woomack]

It's or voltage or temps issue. When I see how my 7700K is acting I can only say that air cooler or something like 240mm AIO is not enough for 1.35V and in most cases not enough for 1.3V+ at full load.
@trents, I don't think this chip requires so high voltage, it should run at 1.20-1.25V at 4.8GHz or something near. Play with LLC settings and maybe vrm frequency if you have these settings. Too high voltage is causing instability ( because of temp spikes ) and 1.35V can be already too high.
Other thing is that maybe you have a hot chip. My 7700K runs at high temps and I'm thinking to delid it this week and test which TIM is the best ( so far Gelid Extreme gives be the best results ).

I still laugh at MSI and their 5.2GHz 1.5V OC profile for AIO water cooling I had to run my 7700K on single stage to deliver low enough temps at 1.5V+. At 1.6V+ I already needed dry ice to see improvements.

 

[trents]

It's or voltage or temps issue. When I see how my 7700K is acting I can only say that air cooler or something like 240mm AIO is not enough for 1.35V and in most cases not enough for 1.3V+ at full load.
@trents, I don't think this chip requires so high voltage, it should run at 1.20-1.25V at 4.8GHz or something near. Play with LLC settings and maybe vrm frequency if you have these settings. Too high voltage is causing instability ( because of temp spikes ) and 1.35V can be already too high.
Other thing is that maybe you have a hot chip. My 7700K runs at high temps and I'm thinking to delid it this week and test which TIM is the best ( so far Gelid Extreme gives be the best results ).

I still laugh at MSI and their 5.2GHz 1.5V OC profile for AIO water cooling I had to run my 7700K on single stage to deliver low enough temps at 1.5V+. At 1.6V+ I already needed dry ice to see improvements.

By the time you subtract the vdroop I'm just about there a 4.8 ghz. This motherboard allows a lot of vdrop and vdroop and I only have 4 levels of LLC to choose from. I've tried them all. I shouldn't have to mess with VRM frequency at this level of overclock I wouldn't think. It's not extreme overclocking.

 

[Woomack]

On MSI I could run my 7700K at 5GHz ~1.275V, on ASUS depends on power options it was between 1.285V and 1.31V. In both cases I was trying to keep load voltage as close to what I set in BIOS so on MSI it was LLC 4-5 and on ASUS it was 5-6. Power options were helping in this case the same as max current limits and some additional settings.
I don't know how it's acting on ASRock but looking at various results on 7700K I just think that your CPU is losing stability because of too high temps caused by too high voltage for your cooling.
Have you tested it making small steps like slowly raising voltage at each ratio ? There should be similar voltage step up to about 4.8-5.0GHz while on most these chips above 5GHz is higher voltage step for next ratio.
I set 1.15V as base voltage and I was able to run it up to 4.7GHz, at 4.8GHz I had to set ~1.18V, at 4.9GHz it was ~1.22V, at 5.0GHz it was ~1.27V. 5.1GHz was already little problem as temps at 1.35V were high even though I have custom loop with 2x 360 rad what for this little core is still not enough. I have in plans delidding my cpu later this week.

 

[trents]

On MSI I could run my 7700K at 5GHz ~1.275V, on ASUS depends on power options it was between 1.285V and 1.31V. In both cases I was trying to keep load voltage as close to what I set in BIOS so on MSI it was LLC 4-5 and on ASUS it was 5-6. Power options were helping in this case the same as max current limits and some additional settings.
I don't know how it's acting on ASRock but looking at various results on 7700K I just think that your CPU is losing stability because of too high temps caused by too high voltage for your cooling.
Have you tested it making small steps like slowly raising voltage at each ratio ? There should be similar voltage step up to about 4.8-5.0GHz while on most these chips above 5GHz is higher voltage step for next ratio.
I set 1.15V as base voltage and I was able to run it up to 4.7GHz, at 4.8GHz I had to set ~1.18V, at 4.9GHz it was ~1.22V, at 5.0GHz it was ~1.27V. 5.1GHz was already little problem as temps at 1.35V were high even though I have custom loop with 2x 360 rad what for this little core is still not enough. I have in plans delidding my cpu later this week.

Yes, I have been testing stability at various voltages and increasing the voltages in small steps until I achieve stability under reasonably demanding stress tests such as 3 hrs. of OCCT large data set and Prime95 27.7 blend for two hours.

Woomack, I'm a little perplexed by your input here. You must be reading entirely different sources than what I have run across for Kaby Lake 7600k/7700k overclocking because what I have been reading indicates that my voltages and temps at various levels of overclock levels are pretty much the norm. In fact, there is a link somewhere on this forum in a recent post that takes you to a website that sells "cherry picked" 7600k and 7700k CPUs. Their criteria for "cherry picked" is that a CPU will pass 1 hr. of Realbench at 1.35 vcore. It takes me 1.36 vcore to pass 1 hr. of the Realbench stress test so I'm not far from that. And "high temps"? Since when does 87c consitute a high temp with Intel? Like I say, it seems you have been reading different stuff than I have. And I have never heard anyone say that temps below Tjmax in themselves create instability. As far as I know, lower temps only improve stability when you get down to really cold like when using LN in extreme overclocking. Am I missing something?

And most veteran overclockers are saying that safe 24/7 voltages for Sky/Kabby is 1.4 or less. What I'm experiencing seems inline with what I'm reading about what others experiencing in overclocking these CPUs. But you must be seeing something different than I am.

 

[EarthDog]

The website is silicon lottery: https://siliconlottery.com

If you will notice, 68% of 7600k's can do 4.9 Ghz 7600k 1.344V. If you are at 4.8 GHz and higher voltage, I'd call that below average considering nearly 70% can do it with slightly less voltage. But I believe you are at 1.32V amd 5ghz.
https://siliconlottery.com/collections/lga-1151/products/7600k49g


7600k amd 7700k percents are different. For example, here is 5 ghz 7700k. 1.37V and almost 60% get there.
https://siliconlottery.com/collections/lga-1151/products/7700k50g

7600k is less than 50%: https://siliconlottery.com/collections/lga-1151/products/7600k50g

 

[trents]

The website is silicon lottery: https://siliconlottery.com

If you will notice, 68% of 7600k's can do 4.9 Ghz 7600k 1.344V. If you are at 4.8 GHz and higher, I'd call that below average considering nearly 70% can do it with slightly less voltage.
https://siliconlottery.com/collections/lga-1151/products/7600k49g

That seems reasonable. So I don't have a cherry but my voltages are in the ballpark at least and I can safely get a decent overclock because my cooling is good. Neither voltages or temps are excessive. And my understanding is that delididng in this case will not increse the max stable overclock.

 

[trents]

But wait a minute, ED! Those stats from Silocon Lottery you refer to are based on being able to pass 1 hr. of the Realbench stress test. I can pass that at 4.8 ghz on definitely less voltage than it takes to pass 2 hr. of Prime95. It seems I need about .02 more vcore to pass Prime than I do Realbench. So that definitely puts me in the ballpark.

 

[trents]

Okay, tested at 4.9 ghz for one hour on the Realbench stress test. Seems like a pretty average CPU I guess. Maybe could have got by with a little lower votage with the primo motherboard Siliconlottery.com used. But will do 5 ghz stable on less than 1.4 vcore and that's what I was hoping for.

 

[R_Pierce]

what ram do you guys recommend for best OC ability? I currently have Corsair Vengeance LPX 3200 DDR4 C14 but anything over the XMP profile (3200) gets unstable and shaky fast. I was wondering what you guys recommend for something that can be overclocked with consistency closer to 38XX or even 4XXX

 

[Alaric]

what ram do you guys recommend for best OC ability? I currently have Corsair Vengeance LPX 3200 DDR4 C14 but anything over the XMP profile (3200) gets unstable and shaky fast. I was wondering what you guys recommend for something that can be overclocked with consistency closer to 38XX or even 4XXX

There are a lot of G. Skill fans here, and there is a reason for that. Here http://www.overclockers.com/forums/showthread.php/732290-Woomack-s-memory-test-list is an excellent place to start.

 

[R_Pierce]

Thinking about this. Its only a couple dollars more than the 3600.

https://www.newegg.com/Product/Product.aspx?Item=N82E16820232463