A suitable vcore for my 8700k @ 4.7GHz

[ClownsBlur_]

Hello to all. I wanted to make this post as I only recently overclocked and want to know roughly what the best suitable vcore would be. Right now I have the vcore to 1.26 and ran Prime95 (small FFTs) for about an hour and no errors and all is good. I monitored temps in core temp and they reported 87-90c max. If anyone else has the same frequency and could let me know their vcore, that'd be nice too. I know all CPU's are different but I'd like to know roughly what the best vcore would be for my frequency.

 

[BTVA]

The best vcore is the lowest one at which the cpu still work perfectly stable. So if you have 1.26v stable, try 1.25v, 1.24v, etc.

 

[mikeo]

As above...

But... There are so many variables at work here. How well your particular CPU Overclocks (every CPU is different), ambient temp, case configuration/cooling, CPU cooler (which you don't mention what you have), thermal paste etc. etc.

As an example. Mine will do 4.7 at 1.2v and 4.9 at 1.28v (both with medium level LLC). This is on a Noctua NH-U14s push/pull in a very well vented aluminium mid tower case, with an ambient room temp of approx. 20C.
Both OC's have been tested stable at 20 runs IBT (high memory use), 1/2 hour Realbench and 6/8hours Prime 95 small ft. Plus of course day-to-day use. Temps are around 65/70c for 4.7 and 70-80c for 4.9 during testing.
Worth noting that every ones definition of a testing strategy tend to vary. But the above has never failed me in some 20 years or so of building PC's. If I should get issues, I then tend to run more tests, though this very rarely happens.

Now going by the above, you would have thought my CPU should hit 5GHz easy. But no it wont, well not on a sensible voltage (and hence temperature). As I said, every CPU is different in what it can... or can not do.

have fun.

PS. personally I would not be happy with temps pushing 90C and would be looking at ways to reduce this (if possible). I know they can hit 100C before starting to throttle, but I personally would not be happy with a day-to-day temp pushing 90C.

 

[mikeo]

As above...

But... There are so many variables at work here. How well your particular CPU Overclocks (every CPU is different), ambient temp, case configuration/cooling, CPU cooler (which you don't mention what you have), thermal paste etc. etc.

As an example. Mine will do 4.7 at 1.2v and 4.9 at 1.28v (both with medium level LLC). This is on a Noctua NH-U14s push/pull in a very well vented aluminium mid tower case, with an ambient room temp of approx. 20C.
Both OC's have been tested stable at 20 runs IBT (high memory use), 1/2 hour Realbench and 6/8hours Prime 95 small ft. Plus of course day-to-day use. Temps are around 65/70c for 4.7 and 70-80c for 4.9 during testing.
Worth noting that every ones definition of a testing strategy tend to vary. But the above has never failed me in some 20 years or so of building PC's. If I should get issues, I then tend to run more tests, though this very rarely happens.

Now going by the above, you would have thought my CPU should hit 5GHz easy. But no it wont, well not on a sensible voltage (and hence temperature). As I said, every CPU if different in what it can... or can not do.

have fun.

PS. personally I would not be happy with temps pushing 90C and would be looking at ways to reduce this (if possible). I know they can hit 100C before starting to throttle, but I personally would not be happy with a day-to-day temp pushing 90C.

EDIT. You also don't explicitly state if you are setting the vcore manually in the BIOS, on auto or allowing the system to run in MCE mode.

 

[ClownsBlur_]

As above...

But... There are so many variables at work here. How well your particular CPU Overclocks (every CPU is different), ambient temp, case configuration/cooling, CPU cooler (which you don't mention what you have), thermal paste etc. etc.

As an example. Mine will do 4.7 at 1.2v and 4.9 at 1.28v (both with medium level LLC). This is on a Noctua NH-U14s push/pull in a very well vented aluminium mid tower case, with an ambient room temp of approx. 20C.
Both OC's have been tested stable at 20 runs IBT (high memory use), 1/2 hour Realbench and 6/8hours Prime 95 small ft. Plus of course day-to-day use. Temps are around 65/70c for 4.7 and 70-80c for 4.9 during testing.
Worth noting that every ones definition of a testing strategy tend to vary. But the above has never failed me in some 20 years or so of building PC's. If I should get issues, I then tend to run more tests, though this very rarely happens.

Now going by the above, you would have thought my CPU should hit 5GHz easy. But no it wont, well not on a sensible voltage (and hence temperature). As I said, every CPU is different in what it can... or can not do.

have fun.

PS. personally I would not be happy with temps pushing 90C and would be looking at ways to reduce this (if possible). I know they can hit 100C before starting to throttle, but I personally would not be happy with a day-to-day temp pushing 90C.

Coolermaster HAF X case, 230mm intake at the front, 140mm outtake (bought my own green LED fans with high CFM (70-80CFM+), Dark Rock 4 CPU cooler (Arctic Silver 5 thermal compound). Also I ran IntelBurnTest on standard, high and very high and all were successful. Temps reached maximum 77c on very high.

 

[mikeo]

Your system looks fine to me

If that's what you are reaching on IBT. Then I'll make a guess and assume you are hitting around 60/65C in gaming (obviously depends on what you are playing). Which is just fine.

I would also hazard a guess if your temps are pushing 90C on Prime95, that you are running one of the later versions that use the AVX instruction set. If so, then be aware that this puts an enormous load on your CPU. Much more so than you are ever likely to encounter day-to-day. And as so, in my opinion (and that of many) not really that good a stress test to run. Though I am aware that some may disagree (as is their right).

Much better to run Prime95 v26.6 and/or something like Realbench. I use a quick run of IBT to establish initial stability and proceed from their. Though some will recommend programs like OCT and many others. And how long to run them? How long is a piece of string? Up to an individual as to what they are happy with. But ultimately, it's a matter of whether it runs stable day-to-day. I've seen people run things like Prime95 for 24 hours and then complain when their PC falls flat on it's face the first game they run.

It's all good fun.

 

[tyler_jrb]

Not the best overclocker however. Absolutely flat out it will do 1.425v @ 5.2ghz delidded running approx 70-75c maximum under 240 AIO.

 

[mikeo]

There's a pretty good article here covering stress testing CPU's:
https://www.tomshardware.com/reviews/stress-test-cpu-pc-guide,5461.html

 

[mikeo]

Not the best overclocker however. Absolutely flat out it will do 1.425v @ 5.2ghz delidded running approx 70-75c maximum under 240 AIO.

Not too bad I would have thought. Looking at the temps, it just goes to show how much the temps are improved by a good de-lidding!
Are they day-to-day running temps, or stress testing?

 

[tyler_jrb]

Not too bad I would have thought. Looking at the temps, it just goes to show how much the temps are improved by a good de-lidding!
Are they day-to-day running temps, or stress testing?

Yeah makes a massive difference. Max temp is 75c with prime95 26.6 on blend with fan speed approx 800-1000rpm.

 

[ClownsBlur_]

Your system looks fine to me

If that's what you are reaching on IBT. Then I'll make a guess and assume you are hitting around 60/65C in gaming (obviously depends on what you are playing). Which is just fine.

I would also hazard a guess if your temps are pushing 90C on Prime95, that you are running one of the later versions that use the AVX instruction set. If so, then be aware that this puts an enormous load on your CPU. Much more so than you are ever likely to encounter day-to-day. And as so, in my opinion (and that of many) not really that good a stress test to run. Though I am aware that some may disagree (as is their right).

Much better to run Prime95 v26.6 and/or something like Realbench. I use a quick run of IBT to establish initial stability and proceed from their. Though some will recommend programs like OCT and many others. And how long to run them? How long is a piece of string? Up to an individual as to what they are happy with. But ultimately, it's a matter of whether it runs stable day-to-day. I've seen people run things like Prime95 for 24 hours and then complain when their PC falls flat on it's face the first game they run.

It's all good fun.

I ran IBT on standard, high and very high, all threads and time to run I set to 5. Is 5 times enough or shall I do 10 (the default when I load up IBT)? Also I have used the latest version of Prime and when running that, my temps were 90-93c and I had a error/fail within 2-3 minutes but on IBT my max temps were 75-77c (very high). I also ran small FFT on Prime95. I had a feeling something was wrong with it. There's no way my temps could be that high especially with a decent cooler.

 

[mikeo]

IBT I would say a min of 10 runs. Mid 70's running this sound about right to me. I know you say that you set the voltage at 1.26 .... But what is it actually pulling? Could well be higher depending on what else settings you've changed etc. In the BIOS. Check with something like hwmonitor.

Try an earlier version of prime 95 and see what that looks like, version 26.6 is probably the best. Consider running something like realbench to simulate more real world loads.

What's your ambient room temp approx?

How did you apply the Tim to your cooler?

Edit. You might be unlucky and the Tim under the heat spreader on your chip has just been applied poorly. Hence why a delid and re applying the Tim etc. can reduce temps in some circumstances so dramatically. Though that's obviously your warranty on the CPU down the toilet if you go this route.

 

[ClownsBlur_]

IBT I would say a min of 10 runs. Mid 70's running this sound about right to me. I know you say that you set the voltage at 1.26 .... But what is it actually pulling? Could well be higher depending on what else settings you've changed etc. In the BIOS. Check with something like hwmonitor.

Try an earlier version of prime 95 and see what that looks like, version 26.6 is probably the best. Consider running something like realbench to simulate more real world loads.

What's your ambient room temp approx?

How did you apply the Tim to your cooler?

Edit. You might be unlucky and the Tim under the heat spreader on your chip has just been applied poorly. Hence why a delid and re applying the Tim etc. can reduce temps in some circumstances so dramatically. Though that's obviously your warranty on the CPU down the toilet if you go this route.

CPU-Z reports 1.26. Ambient right now is 21c - 20-23c mostly. I used the pea method on the CPU then placed cooler on top and screwed down each corner by a little then screwed the other diagonally a little then repeated until all was down tight. I don't want to delid by the way haha.

 

[BTVA]

cpuz only read voltage with a step of 0.016v, so the next lower vcore it could read is 1.248v and 1.232v and so on. Try to lower the voltage in bios first, and see how the cpu behaves in cpuz, that is what I would suggest.

 

[ClownsBlur_]

cpuz only read voltage with a step of 0.016v, so the next lower vcore it could read is 1.248v and 1.232v and so on. Try to lower the voltage in bios first, and see how the cpu behaves in cpuz, that is what I would suggest.

Yeah I notice after a little the voltage dips a little then sticks to 1.26 mostly. I haven't got round to lowering the voltage by -1 and running IBT yet haha. If I can get 1.24 or 1.25 stable @ 4.7GHz then that'd be nice.