Sign in to follow this  
martin-w

6700K - Manual Overclocking Voltage.

Recommended Posts

Just in case anyone's interested, I thought I'd post an update on my overclocking endeavours with the 6700K, and enquire what sort of voltages you chaps are running with.

 

I noticed that the Asus Five Way Optimisation software nudged the adaptive voltage up to 1.42 volts, running BF4. Not excessive, no issue, Asus and Intel tell us that 1.45 is acceptable, but I was pretty sure it was a tad excessive for my chip. Temp running RealBench was 83 degrees

 

So, I dived into some manual overclocking in order to see if I could achieve a stable overclock at lower voltage and as a comparison with the 5WO results.

 

Overclocking manually in the UEFI, I achieved a stable 4.6 GHz overclock, with manual voltage at 1.34 volts. That would be borderline of course, just stable, and it's always a good idea to add 0.01 - 0.02 to the voltage as a margin of error. Therefore I set the adaptive voltage to 1.36 Volts. Under load I now see 1.365 volts and in Real Bench no higher than 75 degrees.

 

The above is with the Noctua NH-D15S, in case anyone's interested in how well it performs with an overclocked 6700K. Fantastic cooler, very quiet, very pleased. Don't let the hype and aggressive marketing sway you towards AIO water coolers before you've thought about it carefully.

 

So I reckon I'd have to agree with those I've discussed this with before on the forum... it does indeed seem that Asus Five Way Optimisation adds more voltage than necessary. Who knows, perhaps Asus are aware of something I'm not and the extra voltage is advantageous. However, so far, all is stable and nice and cool under load. 

 

I's be interested in the voltage those of you with overclocked 6700K's are running with. Is my voltage average, above average or low? What's your voltage? Which cooler are you using and temps in RealBench? 

Share this post


Link to post
Share on other sites
Help AVSIM continue to serve you!
Please donate today!

if you plan to run 24/7 1,45 is to much if you ask me.

Who said? Intel are the ones to ask, they made the CPU... and they say 1.45 is a safe voltage. Asus agree. As long as you have an adequate cooling solution to handle it of course. Many run at up to 1.45 volts, not an issue for Skylake. The Intel spec sheet states a max voltage of 1.5.

 

 

My 6700k runs at 4.6ghz and 1.3v

Is that adaptive, offset or manual voltage? Do you know the temp in RealBench or Cinebench?

Share this post


Link to post
Share on other sites

Is that adaptive, offset or manual voltage? Do you know the temp in RealBench or Cinebench?

 

No idea.  I'm using a Gigabyte Gaming K3, LLC is set to high, I just followed this guide https://forums.overclockers.co.uk/showthread.php?t=18690667 and it worked perfectly.  Only stress testing I did was numerous runs of 3dmark firestrike, and actually gaming.  I've clocked previous rigs meticulously in the past, running AIDA, and Superpi etc, last two I've just run quick tests, then cracked on gaming.

Share this post


Link to post
Share on other sites

I have my i7 6700K OCed to 4.5Ghz at 1.32V I have the Cheepo  Evo 212 Air cooler and its stable

 

Running P3D I get average 65-75C

 

but when loading P3D at the mark...6% loading terrain, all cores and threads are utilized and it reaches max of 87C.

 

Not sure if I can push it and ignore the loading spike part...

Share this post


Link to post
Share on other sites

No idea.  I'm using a Gigabyte Gaming K3, LLC is set to high, I just followed this guide https://forums.overclockers.co.uk/showthread.php?t=18690667 and it worked perfectly.  Only stress testing I did was numerous runs of 3dmark firestrike, and actually gaming.  I've clocked previous rigs meticulously in the past, running AIDA, and Superpi etc, last two I've just run quick tests, then cracked on gaming.

 

 

 

Just had as quick glance at you guide. Looks like it's setting a fixed voltage, with all power saving features disabled.

 

If you're happy with that, fair enough. Using power unnecessarily though. Not that it will break the bank.

 

If you have just set the voltage to 1.3, without verifying that it's a voltage that your specific board requires, just copying the guide... it could well be that your voltage is borderline stable. Conversely, it could be that you could be stable with less. If it fails some time in the future, the first thing to check perhaps.

I have my i7 6700K OCed to 4.5Ghz at 1.32V I have the Cheepo  Evo 212 Air cooler and its stable

 

Running P3D I get average 65-75C

 

but when loading P3D at the mark...6% loading terrain, all cores and threads are utilized and it reaches max of 87C.

 

Not sure if I can push it and ignore the loading spike part...

 

 

 

4.5 is about what I'd expect with the 212 Evo to be honest. 

 

Not sure if I can push it and ignore the loading spike part...

 

No, I wouldn't! If it peaks at 87, leave be. TJMax for Skylake is 105C. Personally I would regard any spike to 87 as the max. Although some are okay with 90's in a vicious stress test.

 

The only time I've seen temps in the mid 80's, with the D15S, is when I ran RealBench, but with the D15S fans still set to PWM and obeying the Asus Fan Xpert 3 standard fan profile. Once I set the CPU fan to the max (still quiet) it was mid 70's.

 

the other point of course is that in terms of frame rate, an extra 100 MHz will give you very little.

 

Same question to you Manny... is that fixed, offset or adaptive voltage. Did you follow a guide?

 

Most straightforward and best guide I've seen is here...

 

https://rog.asus.com/19262015/overclocking/guide-overclocking-core-i7-6700k-on-the-maximus-viii-extreme/

Share this post


Link to post
Share on other sites

Thanks Martin for the update. Just went Skylake as well. Still playing with OC and voltage. Turns out that the RAM sub-timings play a huge role in synthetic benchmark results. Wonder if they translate to RW results...

 

Also, everyone should check out the Rockit88 delid kit. Not affiliated in any way; just placed an order. For those of us that have delidded using either the hammer/vice or razor (for me, the latter), a tool like this is a welcome departure. Lots of tools coming online for this. Cool that folks can make some money like this but I suppose if Intel would just use better TIM for 2 cents a CPU, it'd make life a lot easier...

Share this post


Link to post
Share on other sites

Just a reminder that comparing cooler cpu temps without quoting

the ambient temp at the same time is pretty meaningless.

 

gb.

Share this post


Link to post
Share on other sites

gboz, on 06 Sept 2016 - 09:43 AM, said:

 

Just a reminder that comparing cooler cpu temps without quoting

the ambient temp at the same time is pretty meaningless.

 

gb.

 

 

Yes, indeed! When comparing with temp results of others. Not required in my first post in this thread, as I wasn't comparing my temp to that of others, only 5WO to my manual overclocking. For reference, ambient was 25 degrees in all cases. In other posts, other threads, I refer to delta temp.

 

Not exactly meaningless though, as comfortable room temperatures for we human beings are within a specific range. But yes certainly not 100% accurate. I'm sure if we delve into accurate comparisons in this thread we will be quoting delta temp. Thanks for reminding us though, it's right that you should.

 

 

garrett_frank, on 06 Sept 2016 - 09:34 AM, said:

 

Thanks Martin for the update. Just went Skylake as well. Still playing with OC and voltage. Turns out that the RAM sub-timings play a huge role in synthetic benchmark results. Wonder if they translate to RW results...

 

Also, everyone should check out the Rockit88 delid kit. Not affiliated in any way; just placed an order. For those of us that have delidded using either the hammer/vice or razor (for me, the latter), a tool like this is a welcome departure. Lots of tools coming online for this. Cool that folks can make some money like this but I suppose if Intel would just use better TIM for 2 cents a CPU, it'd make life a lot easier...

 

Hi Garett.

 

I have seen the delid tools. To be honest I have no desire to delid at the moment. In my opinion it's not worth it for me. I'm at 4.6 GHz currently, with a reasonable thermal and voltage margin to push higher. Won't for now though, as in my opinion the extra 100 or maybe 200 MHz equates to very little in terms of frame rate. Overclocking in a well balanced system is linear as I'm sure you know, thus 1 or 2 frames per second at 30 frame per second. Meaningless in terms of frame rate.

 

I think if you have a chip that generates a lot of heat though it's worth it. Either as a result of Intel's IHS mounting being dodgy or simply as a result of being less favoured by the silicone lottery.

 

If I change my mind, it will be for the hell of it, for the challenge, for fun. :smile:

 

 

Quotebut I suppose if Intel would just use better TIM for 2 cents a CPU, it'd make life a lot easier...

 

 

It's not so much the quality of Intel's TIM, it's actually reasonable quality. The issue is more to do with the adhesive that attaches the IHS to the substrate. The band of adhesive around the IHS, increases the gap with the die, thus more TIM required to fill the void. Thick TIM equates to higher temps.

 

Liquid Pro or Ultra is thinner, but yes, superior in terms of thermals too.

 

Regarding Intel's motivation for switching to TIM instead of solder, it actually makes sense to be honest, for a number of reasons.

 

The IHS is copper, die is silicone. Two entirely different materials. Tin based solder won't stick to silicone. Tin also shrinks by a huge factor which increase thermal tension. Thus the only choice is Indium solder. Indium is as a very rare material and very expensive. The IHS needs to be coated with a Nickel layer to allow the Indium solder to stick. Still doesn't stick that well so It requires another layer, Palladium Silver or Gold. And I don't need to tell you how expensive they are.

 

It doesn't end there, the process gets even more complex. Soldering an IHS to a die is time consuming and expensive.

 

Lots of other issues to circumvent, like deformed heat spreaders, micro cracks. Voids and micro cracks are two of the main failure mechanisms that pushed Intel toward TIM instead of solder for small die CPU's like Skylake.

Share this post


Link to post
Share on other sites

Not exactly meaningless though, as comfortable room temperatures for we human beings are within a specific range. But yes certainly not 100% accurate. I'm sure if we delve into accurate comparisons in this thread we will be quoting delta temp. Thanks for reminding us though, it's right that you should.

 

 

Yes, agree, "much more meaningful" is what I should of said.

 

 

I's be interested in the voltage those of you with overclocked 6700K's are running with. Is my voltage average, above average or low? What's your voltage? Which cooler are you using and temps in

 

 

Just monitored a P3D session with HWMonitor and with my 5WO OC the volts spent most of the time around 1.37 with frequent drops to about 1.2 and a max to about 1.42.

 

gb.

Share this post


Link to post
Share on other sites

Yes, agree, "much more meaningful" is what I should of said.

 

 

 

 

Just monitored a P3D session with HWMonitor and with my 5WO OC the volts spent most of the time around 1.37 with frequent drops to about 1.2 and a max to about 1.42.

 

gb.

 

 

 

Thanks, that's what I was seeing with 5WO, OC 4.6. Peaking to around 1.42. 5WO sets adaptive voltage of course.  I was seeing 83 degrees, ambient 25C, so 58 Delta.

 

Manually overclocking I'm now seeing peaks to 1.36 and lower temp. 75C in RealBench. So 50C Delta.

 

Just ran Aida 64. Without FPU, so no AVX, 68C. With FPU so running AVX, still only 81C. Pretty good for AVX.

 

And I should add, importantly, that temp was with the Fan Xpert 3 profile I configured, so CPU fans and case fans were only running at 50% with the zero FPU run. Closer to max with the FPU run.

 

Certainly seems the case that 5WO over volts to a degree. 

Share this post


Link to post
Share on other sites

Just had as quick glance at you guide. Looks like it's setting a fixed voltage, with all power saving features disabled.

 

If you're happy with that, fair enough. Using power unnecessarily though. Not that it will break the bank.

 

If you have just set the voltage to 1.3, without verifying that it's a voltage that your specific board requires, just copying the guide... it could well be that your voltage is borderline stable. Conversely, it could be that you could be stable with less. If it fails some time in the future, the first thing to check perhaps.

 

Yeh, the PC is only on when I'm using it, so no problem there.  I've had extended gaming sessions on it with no issues.  Might tweak it when I have a chance, but no need at the minute.

Share this post


Link to post
Share on other sites

I've just got hold of a skylake 6700k with a gigabyte gaming 7 board. I'm using the nh d14 as the 15 wouldn't fit in my case.

Not really got any idea about overclocking but I've read that using the auto overckock isn't a good idea.

Is this right?

Share this post


Link to post
Share on other sites

I have my 6700K running at 4.80Ghz HT Off, 24/7 at 1.42 volts and on my Corsiar H110GTX and never break 78c under full load. This is also in conjunction with my XMP Profile with the DDR4 3200Mhz memory. 

Share this post


Link to post
Share on other sites

Hi Martin

Here is some very nice tests from Roman

http://overclocking.guide/skylake-overclocking-power-consumption-and-voltage-scaling/

 

For 24/7 on custom watercoling i prefer to have 1.42v or less on air 1.32-1.35 with HT on,

With HT off 1.45 custom water and 1.35-1.4v air.

Have one 6700k that do 5.1ghz HT off 1.43 with H110 stable Max temp in p3d 70C , HT on 5.0 1.39v

 

Ramspeed do approx 0.8% each 100mhz go from c16 to c15 approx 0.6% in P3D v3.3.2

I have a nice set tridentZ 3600 cl15 kit run 3600 cl13 1.4v 24/7.

Share this post


Link to post
Share on other sites

 

 


Just a reminder that comparing cooler cpu temps without quoting

the ambient temp at the same time is pretty meaningless.

 

I saw a comment on a youtube video (about delidding) bragging about how someone had gotten their temps down to 10*C. They're either keeping it in their mini-fridge or outside in the cold. Then again, it was YT we're talking about...

Share this post


Link to post
Share on other sites

I've just got hold of a skylake 6700k with a gigabyte gaming 7 board. I'm using the nh d14 as the 15 wouldn't fit in my case.

Not really got any idea about overclocking but I've read that using the auto overckock isn't a good idea.

Is this right?

 

Try this, should be the same BIOS settings with it being Gigabyte  https://forums.overclockers.co.uk/showthread.php?t=18690667

  • Upvote 1

Share this post


Link to post
Share on other sites

Not really got any idea about overclocking but I've read that using the auto overckock isn't a good idea.

Is this right?

 

 

 

Well no, that's not true. I have no experience of Gigabyte boards but in terms of the Asus equivalent, 5 Way Optimisation, that's certainly not the case. Gigabyte are a good manufacturer, so I doubt it's the case for their software either.

 

You see this a lot in forums, as soon as you mention auto overclocking they throw their arms in the air, wring their hands and warn of death and destruction, exploding motherboards and thermonuclear meltdown.  :smile:

 

These individuals just copy each other to be honest. One self appointed guru or another claims auto overclocking is bad [may have been, in the context he was referring to] and all the other lemmings jump on the bandwagon and pass on the same information, whether it's actually true for that specific software or not.

 

Yes indeed, from time to time you come across auto overclock software that's in need of a patch or two, and yes, perhaps not perfected and best avoided but to generalise and condemn all auto overclocking software from all manufactures is stupidity.

 

What I was referring to in this thread was the Asus Five Way Optimisation auto overclocking software. I mention how it adds a bit more voltage than when I manually overclock. It's not  a vast increase in voltage, it's not disastrous, it's not beyond a safe level. In fact, 5WO is designed to only increase the voltage within safe levels. 

 

5WO also includes the facility to set a max voltage, max temperature or max frequency. So the user has total control over the process. It's a perfectly valid approach to use 5WO to overclock, and then nip into the UEFI and tweak the voltage down if you desire. That way, using 5WO as a starting point.

 

As usual I've babbled on needlessly, but NO, it's not true that all auto overclocking software should be avoided and is a pile of doggy poo. In fact quite a few on this forum and some in this thread, overclock using 5WO.

 

I have my 6700K running at 4.80Ghz HT Off, 24/7 at 1.42 volts and on my Corsiar H110GTX and never break 78c under full load. This is also in conjunction with my XMP Profile with the DDR4 3200Mhz memory. 

 

 

What kind of full load Angelo? Stress test? If so which one, and AVX or not? The H110 is a great cooler, so I suspect you must mean in a pretty tough stress test. Ambient? 

 

 

, have delidded all my 6700k CPU.s replaced the TIM with Kryonaut Grizzly.

 

https://www.dropbox.com/s/y4f9q4p5drp3t83/48505.png?dl=0

 

 

 

 

Hi Hasse.

 

 Why did you go for Grissly instead of Liquid Pro or Ultra. What kind of temp reduction from delliding did you see?  :smile:

Share this post


Link to post
Share on other sites

Just tried them settings and everything is running ok at the moment.

Haven't stressed test it properly yet, any advice on the best program for that.

I've downloaded cpu z but apparently it's not very good for testing.

 

I tend not to bother any more.  Run a few consecutive 3dmark benchmark runs, and if it's stable, I start gaming with it.  A few extended sessions, if it hasn't fallen over, it's unlikely to.  Obviously just watch the temperatures under load

Share this post


Link to post
Share on other sites

 

 

 

What kind of full load Angelo? Stress test? If so which one, and AVX or not? The H110 is a great cooler, so I suspect you must mean in a pretty tough stress test. Ambient? 

 

 

 

 

I used OCCT various data sets all 4 cores pinned 100% load for hours. After that I just ran various FSX and P3D scenarios for my real world testing. It all passed and checked out. Ambient air temp is 72F or 22C. 

Share this post


Link to post
Share on other sites

Hi Martin

Have a lot of Grizzly , and Its the only that work if a want to go zubzero.

 

Tested one normal 6700k for fun think the grizzly do 6-10c not sure.

 

Did a short video with bios settings , Aida64 one hour test at 4.6ghz and Test in P3D v3.3.2.

4.6 HT on AM116 , 4.6 HT off no AM and 4.8 HT off.

 

Share this post


Link to post
Share on other sites
Have a lot of Grizzly , and Its the only that work if a want to go zubzero.

 

 

Are you still tinkering with sub zero then Hasse?

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this