i7 6700k OC advice

[eduard]

Hi, I haven't overclocked anything before so I could used some advice.

 

I have a i7-6700K CPU and solid build all around. Have overclocked it to 4.6GHz and got it stable with 1.375V. Max temp underload is below 70*C and I have a NH-D15 cooler.

 

XMP is enabled with memory at 3000MHz.

 

Should I try to push it more? Is the voltage low/high? What are your thoughts?

[Gonio]

Don't go over 1.4V is recommended in general, so it seems you have some more room you can push it. Other then that I think with current voltage you can already crank the multiplier up a few times.

[eduard]

@Gonio If I set the multiplier to 47 with current voltage, it crashes when I start Cinebench.  Am thinking this particular one is not a good overclocker from what I've read.

[Master Disaster]

With that cooler you should be OK for 1.4v but I wouldn't go beyond that.

 

If your 6700K is anything like mine jumping from 4.6 to 4.7 will require a huge voltage boost and will be out of your reach on air.

 

Remember that mid 80s C is the absolute max you'd ever want to see under reasonable load, by that I mean Prime95 Torture Test will obviously push beyond that but using P95 mixed test or something like AIDA or OCCT should not exceed it.

[Simon771]

I think you should just stick with that OC.

Do you acctually need more performance that it can offer you right now?

You are near 1,4V and you should avoid higher voltages, to be safe.

 

For 24/7 use I would't even go higher than 1,35V.

[belfouf]

21 minutes ago, eduard said:

Hi, I haven't overclocked anything before so I could used some advice.

 

I have a i7-6700K CPU and solid build all around. Have overclocked it to 4.6GHz and got it stable with 1.375V. Max temp underload is below 70*C and I have a NH-D15 cooler.

 

XMP is enabled with memory at 3000MHz.

 

Should I try to push it more? Is the voltage low/high? What are your thoughts?

 

[eduard]

@belfouf Thank you for this chart. It's quite informative.

 

Can you detail a bit on what Uncore/Cache/Ring ratio is and what they affect?

[Master Disaster]

2 minutes ago, eduard said:

@belfouf Thank you for this chart. It's quite informative.

 

Can you detail a bit on what Uncore/Cache/Ring ratio is and what they affect?

Its the effective clock speed the Cache bus is running at. Back in the Haswell days overclocking the cache had some effect but since Skylake its been repeatedly shown that OCing cache does nothing to improve performance and only generates additional heat.

 

Just allow your board to control the Uncore Ratio.

[belfouf]

54 minutes ago, eduard said:

@belfouf Thank you for this chart. It's quite informative.

 

Can you detail a bit on what Uncore/Cache/Ring ratio is and what they affect?

Quote

Its the effective clock speed the Cache bus is running at. Back in the Haswell days overclocking the cache had some effect but since Skylake its been repeatedly shown that OCing cache does nothing to improve performance and only generates additional heat.

 

Just allow your board to control the Uncore Ratio.

Indeed, don't push Uncore ratio to the same extent as Core. My i5 6600K has 4.3Ghz Uncore 4.8Ghz Core stable but 4.4Ghz Uncore 4.6Ghz Core unstable.

Really, focus on Core and forget about Uncore

 

 

[done12many2]

2 hours ago, Master Disaster said:

Its the effective clock speed the Cache bus is running at. Back in the Haswell days overclocking the cache had some effect but since Skylake its been repeatedly shown that OCing cache does nothing to improve performance and only generates additional heat.

 

Just allow your board to control the Uncore Ratio.

 

Actually, CPU cache does still impact CPU intensive workloads in both Skylake and Kaby Lake.  Saying something does nothing is a bit of an overreach.

 

2 hours ago, belfouf said:

Indeed, don't push Uncore ratio to the same extent as Core. My i5 6600K has 4.3Ghz Uncore 4.8Ghz Core stable but 4.4Ghz Uncore 4.6Ghz Core unstable.

Really, focus on Core and forget about Uncore

 

 

 

Agreed the CPU clock trumps all, but in some cases, cache speed can stabilize the CPU core at high clock rates.

 

4 hours ago, eduard said:

@Gonio If I set the multiplier to 47 with current voltage, it crashes when I start Cinebench.  Am thinking this particular one is not a good overclocker from what I've read.

 

What motherboard do you have and what is your LLC set to?

 

When chasing your initial overclock, do so with XMP off.  Your goal is to get the highest CPU clock speed and CPU cache speed that you can.  Then worry about memory on the back end.

[Master Disaster]

17 minutes ago, done12many2 said:

Actually, CPU cache does still impact CPU intensive workloads in both Skylake and Kaby Lake.  Saying something does nothing is a bit of an overreach.

 

A 100mhz decrease in clock speed has a much bigger affect on performance than a 1000mhz decrease in cache frequency.

 

With Skylake overclocking Cache is totally pointless.

[done12many2]

5 minutes ago, Master Disaster said:

 

A 100mhz decrease in clock speed has a much bigger affect on performance than a 1000mhz decrease in cache frequency.

 

With Skylake overclocking Cache is totally pointless.

 

CPU cache helps out on CPU intensive applications.  I'm only telling you what I and most folks consistently benchmarking already know.  CPU clock speed will always trump all and I believe I said this in my previous post,  but to tell the OP that CPU cache "does nothing" is far from the truth.

 

Additionally, CPU cache speed has influence on stability in some applications.

 

 

 

[Master Disaster]

53 minutes ago, done12many2 said:

 

CPU cache helps out on CPU intensive applications.  I'm only telling you what I and most folks consistently benchmarking already know.  CPU clock speed will always trump all and I believe I said this in my previous post,  but to tell the OP that CPU cache "does nothing" is far from the truth.

 

Additionally, CPU cache speed has influence on stability in some applications.

 

 

 

I don't think I said it does nothing though,  what I said was it does nothing to improve performance which is a statement I fully stand by.

 

Pushing uncore from stock to 4.6Ghz on my 6700K has literally zero impact on performance in any synthetic benchmark I tried including Cinebench (which is a straight up CPU benchmark), infact in 3D Mark it made my score worse. All it did was add around 3°c to my package temp on my CPU.

 

This result is confirmed in the multiple Skylake benchmark guides available around the major tech sites, overclocking cache does not increase performance in any meaningful way.

[smokefest]

my 6700k runs @ 4.5 Ghz @ 1.335 Vcore with 32 GB ram @ 3000 mhz @ XMP   with a heatsink coolermaster evo 212

 

runs pretty smooth, stable,

In synthetic benchmark I get temp as high as 83-84 c

In real life gaming , like after a full 12 hours of gaming in battlefield 1 which is very cpu intense,

Temps get as high as 75-76-77-78c     ( I check at the end of the day before shutting down my PC in real temp, max temp of the day is often close to 77-78 ) 

Sometimes stays under 75... msot of the time highest temp of the day was 77-78

 

 

What you guys think ? too hot ? all good ? any feed backs ?

[done12many2]

1 hour ago, Master Disaster said:

I don't think I said it does nothing though,  what I said was it does nothing to improve performance which is a statement I fully stand by.

 

Pushing uncore from stock to 4.6Ghz on my 6700K has literally zero impact on performance in any synthetic benchmark I tried including Cinebench (which is a straight up CPU benchmark), infact in 3D Mark it made my score worse. All it did was add around 3°c to my package temp on my CPU.

 

This result is confirmed in the multiple Skylake benchmark guides available around the major tech sites, overclocking cache does not increase performance in any meaningful way.

 

We're going to have to simply agree to disagree.  

 

I'll check out your Cinebench R15 scores in the LTT Cinebench thread.  Hopefully I see you near the top of your bracket for any given clock speed.  

 

Personally, I will continue to overclock cache as I know that contrary to what you're stating, it DOES impact Cinebench R15 scores and it DOES impact CPU intense applications.  I'm not saying that it's profound, but it is definitely measurable.

 

@Jumper118

@Lays

[Jumper118]

8 minutes ago, done12many2 said:

 

We're going to have to simply agree to disagree.  

 

I'll check out your Cinebench R15 scores in the LTT Cinebench thread.  Hopefully I see you near the top of your bracket for any given clock speed.  

 

Personally, I will continue to overclock cache as I know that contrary to what you're stating, it DOES impact Cinebench R15 scores and it DOES impact CPU intense applications.  I'm not saying that it's profound, but it is definitely measurable.

 

@Jumper118

@Lays

it makes a difference. 

[Master Disaster]

48 minutes ago, done12many2 said:

 

We're going to have to simply agree to disagree.  

 

I'll check out your Cinebench R15 scores in the LTT Cinebench thread.  Hopefully I see you near the top of your bracket for any given clock speed.  

 

Personally, I will continue to overclock cache as I know that contrary to what you're stating, it DOES impact Cinebench R15 scores and it DOES impact CPU intense applications.  I'm not saying that it's profound, but it is definitely measurable.

 

@Jumper118

@Lays

See the section called Cache Frequency Doesn't matter for a more than full suite of benchmarks with full testing methodology explained.

 

http://www.overclock.net/t/1570313/skylake-overclocking-guide-with-statistics

 

There's no disagreement to be had, cache frequency doesn't matter on Skylake and these benchmarks prove it categorically.

[Lays]

2 minutes ago, Master Disaster said:

See the section called Cache Frequency Doesn't matter but a more than full suite of benchmarks with full testing methodology explained.

 

http://www.overclock.net/t/1570313/skylake-overclocking-guide-with-statistics

 

There's no disagreement to be had, cache frequency doesn't matter on Skylake and these benchmarks prove it categorically.

If you don't have to sacrifice core freq to turn your cache up a little, there's no harm no foul.  The difference isn't massive when OC'ing cache, but it is there, albeit tiny.

[Master Disaster]

2 minutes ago, Lays said:

If you don't have to sacrifice core freq to turn your cache up a little, there's no harm no foul.  The difference isn't massive when OC'ing cache, but it is there, albeit tiny.

The actual ratio is a 0.7Ghz Cache OC equals around a 0.05Ghz Core OC.

[Lays]

1 minute ago, Master Disaster said:

The actual ratio is a 0.7Ghz Cache OC equals around a 0.05Ghz Core OC.

Like I said, if you don't have to sacrifice any core OC, it's fine.

[Master Disaster]

3 minutes ago, Lays said:

Like I said, if you don't have to sacrifice any core OC, it's fine.

In my opinion its not worth the extra heat tradeoff for such a small gain in performance but hey, each to their own and all that. No one died and made me the OC god.

[belfouf]

On 16/01/2017 at 9:53 PM, smokefest said:

my 6700k runs @ 4.5 Ghz @ 1.335 Vcore with 32 GB ram @ 3000 mhz @ XMP   with a heatsink coolermaster evo 212

 

runs pretty smooth, stable,

In synthetic benchmark I get temp as high as 83-84 c

In real life gaming , like after a full 12 hours of gaming in battlefield 1 which is very cpu intense,

Temps get as high as 75-76-77-78c     ( I check at the end of the day before shutting down my PC in real temp, max temp of the day is often close to 77-78 ) 

Sometimes stays under 75... msot of the time highest temp of the day was 77-78

 

 

What you guys think ? too hot ? all good ? any feed backs ?

I would delid. I did it to my 6600k went from 70°C Max to 50°C Max

Just bought 6700K for my photowork, will delid too

[smokefest]

8 hours ago, belfouf said:

I would delid. I did it to my 6600k went from 70°C Max to 50°C Max

Just bought 6700K for my photowork, will delid too

Don't feel like its worth taking the risk to delid just to drop 10c ( most people get 10c )

 

[done12many2]

1 minute ago, smokefest said:

Don't feel like its worth taking the risk to delid just to drop 10c ( most people get 10c )

 

 

Is that in your own experience?

 

I can tell you that in my own experience, it a lot more than 10c.  Delid only can net you a 15c + reduction alone and when combined with something like CLU, 20c+.

[smokefest]

1 minute ago, done12many2 said:

 

Is that in your own experience?

 

I can tell you that in my own experience, it a lot more than 10c.  Delid only can net you a 15c + reduction alone and when combined with something like CLU, 20c+.

i mean from all the post ive seen online.

Anyway, im not gonna take the risk feel like its not worth it