8700K @5GHz 1.28V

[VicciVicVic]

I managed to achieve exactly what the title says. I haven't seen to many results from other people so I just wanted to know if this is good or bad or average. I know the 8700K can hit 5GHz pretty easily but I don't really know if the voltage is good. I am also on air running less than 80C.

[GoldenLag]

Wait...... And thats stable under load?

 

You could reach 5.2 with that one then at a higher voltage. 

 

I guess you won the jackpot in the silicon lottery

[Not Normie]

yeah try 1.30v and 5.1ghz

 

 

[Jurrunio]

it's very good, though I have seen 5GHz 1.25V so it's not the best

[Not Normie]

yeah my 8600k 5ghz is stable at 1.25v

[DarkSmith2]

1 hour ago, VicciVicVic said:

I managed to achieve exactly what the title says. I haven't seen to many results from other people so I just wanted to know if this is good or bad or average. I know the 8700K can hit 5GHz pretty easily but I don't really know if the voltage is good. I am also on air running less than 80C.

its above average, but "good" is relative, some people can do 5GHz avx stable @1.2v 

But most owners of those are binning CPUs as a hobby, they buy, test and then resell if its not good enough, sometimes with delidding services for little money.

So in the end they pay alot to get such a CPU and i think getting a CPU thats doing 5GHz below 1.3v is a Jackpot scenario for nomal consumers, that didnt payed more or dont have special sources to vendors to get older Batches of the CPU.

1 hour ago, Not Normie said:

yeah try 1.30v and 5.1ghz

probably wont work, Coffeelake scaling is normally between 0.04v-0.06v per 100MHz. But if he is lucky he can aim for 5.2 @1.36v just need delidding and maybe better cooling.

[Not Normie]

worked on my 8600k

 

[DarkSmith2]

3 minutes ago, Not Normie said:

worked on my 8600k

 

sure but you said you only need 1.25v for 5GHz, its logical that you only need 1.30v for 5.1  Nice Chip btw!

[Not Normie]

yeah yeah he probably shouldn't push any further due to cooling anyway.

[DarkSmith2]

Just now, Not Normie said:

yeah yeah he probably shouldn't push any further due to cooling anyway.

yea temperatures always suck with those coffeelake suckers.. If it would be better i would run mine 5.2GHz @1.33v 24/7.. just dont like temperatures above 70°C. I know it would be fine but i dont like them.. Gonna go full custom loop and then we'll see.

[Not Normie]

yeah the tim is pretty bad on intels cpus 

 

 

[DarkSmith2]

6 minutes ago, Not Normie said:

yeah the tim is pretty bad on intels cpus 

oh i delidded a while ago, thats not the problem anymore  See my before and after temps:

[done12many2]

Comparing clockspeeds and voltages from 2 completely different setups operated in different environments with different cooling is a complete waste of time.  Not only that, but each of you test stability completely differently.  

 

One guy just found a better way to show that his CPU runs at a lower voltage.  

[Not Normie]

damn i need to delid temps get up to 70c cant even push 1.35v without hitting 90c

[DarkSmith2]

5 minutes ago, done12many2 said:

Comparing clockspeeds and voltages from 2 completely different setups operated in different environments with different cooling is a complete waste of time.  Not only that, but they way each of you test stability is completely different.  

 

One guy just found a better way to show that his CPU runs at a lower voltage.  

Its fun to do so, cant be a waste  But i think the least people will do for stability testing is AiDA and the most is Prime95 with AVX, so differences should be within 50mV

5 minutes ago, Not Normie said:

damn i need to delid temps get up to 70c 

Hell yea, if you dont give a F*** about warranty and are comfortable enough to do it, because there is a risk to kill it, there is no reason not to.

[done12many2]

6 minutes ago, done12many2 said:

Comparing clockspeeds and voltages from 2 completely different setups operated in different environments with different cooling is a complete waste of time.  Not only that, but each of you test stability completely differently.  

 

One guy just found a better way to show that his CPU runs at a lower voltage.  

1 minute ago, DarkSmith2 said:

Its fun to do so, cant be a waste  But i think the least people will do for stability testing is AiDA and the most is Prime95 with AVX, so differences should be within 50mV

 

 

See the problem with that is that in addition to the reasons I already listed, other factors like LLC used, quality of board, memory overclock, cache overclock and a great deal more play into the comparisons.  Some people intentionally overshoot LLC just to falsely show a lower VCore in a no or light load situation.  

 

One guy bragging that he is running 5 GHz at 1.25v with stock or XMP settings on his RAM and stock cache speed is not even comparable to another guy doing 5 GHz at 1.25v with super tight timings on his RAM and 5 GHz on the cache.   There would be a night and day performance difference between the 2 overclocks compared meaning that the load on the 2nd overclock example is far greater during stress testing than the first example.  The 2nd example would be a far greater chip.

 

Cooling alone will dramatically impact voltage requirements on any chip.  Fact, a cooler chip will require less voltage.  

 

You're absolutely right though, it is fun which is why everyone keeps doing it despite it being pointless.  

[Not Normie]

what tool should i use?

[done12many2]

3 minutes ago, Not Normie said:

what tool should i use?

 

Rockit makes a great and inexpensive tool.  I have the Rockit tool for 3 different socket/cpu types and all of them work flawlessly.

 

Good luck man.  Delidding is great. 

 

https://rockitcool.myshopify.com/

[Not Normie]

yeah, llc helps a lot.

[Velcade]

If my temps stay below 80'C running prime95 for 3hrs is there any improved life expectancy of the CPU when the voltage is lower?  Obviously it should drop temps slightly but will it make that much of a difference?

[DarkSmith2]

46 minutes ago, done12many2 said:

 

 

See the problem with that is that in addition to the reasons I already listed, other factors like LLC used, quality of board, memory overclock, cache overclock and a great deal more play into the comparisons.  Some people intentionally overshoot LLC just to falsely show a lower VCore in a no or light load situation.  

 

One guy bragging that he is running 5 GHz at 1.25v with stock or XMP settings on his RAM and stock cache speed is not even comparable to another guy doing 5 GHz at 1.25v with super tight timings on his RAM and 5 GHz on the cache.   There would be a night and day performance difference between the 2 overclocks compared meaning that the load on the 2nd overclock example is far greater during stress testing than the first example.  The 2nd example would be a far greater chip.

 

Cooling alone will dramatically impact voltage requirements on any chip.  Fact, a cooler chip will require less voltage.  

 

You're absolutely right though, it is fun which is why everyone keeps doing it despite it being pointless.  

im well aware  And im proud to say im testing with Prime95 AVX 80/80 small FFT's 1hour+ (realized if it can do it, it should do the full 23hr test) and have my vcore set to 1.235 in bios resulting in 1.248v in Windows including my level 6 LLC with 4500MHz cache and using an Alphacool Eisbaer LT360 with Nocuta Chromax black.swap Fans  

Just started to fiddle with RAM though, but with the overclocks ive achieved there wasnt any difference in CPU stability yet. Highest ive been was 3866 CL17 or 3600 CL14 though. Using the popular VCCIO/SA settings of 1.10v and 1.15v. Do you need more information, to accept the vcore specification in my signature as valid?

[done12many2]

24 minutes ago, DarkSmith2 said:

im well aware  And im proud to say im testing with Prime95 AVX 80/80 small FFT's 1hour+ (realized if it can do it, it should do the full 23hr test) and have my vcore set to 1.235 in bios resulting in 1.248v in Windows including my level 6 LLC with 4500MHz cache and an Alphacool Eisbaer LT360 with Nocuta Chromax black.swap Fans  

 

I personally would never assume that because an overclock can do one small range for over an hour, that it will automatically pass 23 hours as I know first hand that's not true.

 

With that said, you are still testing stability a great deal harder than most people do and I applaud you for that.  

 

Quote

Just started to fiddle with RAM though, but with the overclocks ive achieved there wasnt any difference in CPU stability yet. Highest ive been was 3866 CL17 though. Using the popular VCCIO/SA settings of 1.10v and 1.15v. Do you need more information, to accept my vcore specification in my signature as valid?

 

As I'm sure you're aware, if you are testing with Prime95 small FFTs, you aren't stressing memory or IMC.  More data across the RAM keeps CPU cycles fed more often, which increases load.  This would be where you'd want to run large FFTs or a long-term 23hr blend test before climbing too high up on your horse about Vcore.  

 

Either way, I'm not questioning your overclock in the least bit nor did I at anytime do so.  I'm sure it's every bit as stable as you need it to be and that is all that matters.

 

If you'll go back through what I said, comparing it to another's overclock is pointless for very obvious reasons.  There are just too many variables from system to system and environment to environment to make any sense out of comparing a known clock to a known voltage. 

 

Showing outputted work though, that's a bit different.  If people want to demonstrate how fast their CPU can accomplish a task, that's quantifiable.  I can guarantee that the guy with a better tweaked overclock despite having a lower clockspeed, will generally beat out the guy with a higher clockspeed, but using stock cache and weak memory settings (generally speaking).

 

49 minutes ago, Velcade said:

If my temps stay below 80'C running prime95 for 3hrs is there any improved life expectancy of the CPU when the voltage is lower?  Obviously it should drop temps slightly but will it make that much of a difference?

 

It will not make that much of a difference in life expectancy.  At least not enough to justify more cooling unless you need more thermal overhead for other things.

 

If you are operating safely while testing with Prime, it is safe to assume that you'll be perfectly fine in any other daily scenario that your CPU will face. 

 

Good luck man.

 

[DarkSmith2]

26 minutes ago, done12many2 said:

I personally would never assume that because an overclock can do one small range for over an hour, that it will automatically pass 23 hours as I know first hand that's not true.

yea to many variables, but i did some reasonable testing which havnt let me down ever since. I mean it depends on how complex you are overclocking, if you include RAM and cache overclocking to the max. then it might not be enough, but for vcore evaluation without any huge extra's it seems to work very well.

 

32 minutes ago, done12many2 said:

If you'll go back through what I said, comparing it to another's overclock is pointless for very obvious reasons.  There are just too many variables from system to system and environment to environment to make any sense out of comparing a known clock to a known voltage. 

Yea it makes sense what you said, never was in doubt. 

32 minutes ago, done12many2 said:

Showing outputted work though, that's a bit different.  If people want to demonstrate how fast their CPU can accomplish a task, that's quantifiable.  I can guarantee that the guy with a better tweaked overclock despite having a lower clockspeed, will generally beat out the guy with a higher clockspeed, but using stock cache and weak memory settings (generally speaking).

Yea im trying to improve my overclocks to be more "rounded" but if you never really did RAM OC its really a pain, and there isnt any guide or poeple to help you, you just have to go through the fiddling with try and error i guess. I have no idea, its really hard to see that some poeple that bought the same Kit are just using Raja's Apex 3700cl17 profile settings bump it to 4133 and add a little voltage, unfortunately i cant, doesnt boot at all. I havent upped cache considerably because i wanted to get my RAMOC better first, but i dont really know what im doing there  

[DildorTheDecent]

3 hours ago, VicciVicVic said:

I haven't seen to many results

Aye...and we haven't seen yours either?

[Velcade]

57 minutes ago, done12many2 said:

It will not make that much of a difference in life expectancy.  At least not enough to justify more cooling unless you need more thermal overhead for other things.

 

If you are operating safely while testing with Prime, it is safe to assume that you'll be perfectly fine in any other daily scenario that your CPU will face. 

 

Good luck man

Thanks!