My i7-7700k won't overclock to anything higher than 4.5ghz

Nov 28, 2018
21
0
10
My specs
motherboard: Msi Z270 sli plus
Cpu: i7-7700k oc (4.5)
Ram: 64gb ddr4 oc (3000)
Gpu: Gtx 1080 ti
Psu: 1000 watt

My Bios Settings:
https://ibb.co/bbVRZpj
https://ibb.co/WfYF9L8
https://ibb.co/WBF2jmk
https://ibb.co/1zXtKPH

I seen lots of people with the same cpu and motherboard and overclock their cpu to 4.8ghz and have it boot to windows and run benchmarks with good temps. Then I try, I changed the settings around making the cpu ratio x46, x47, and x48 with core voltage at 1.280. Only x46 was able to boot and run a benchmark, then after an hour it crashed and I had to lower it back. both x47 and x48 weren't able to boot they either freeze at the when loading windows or windows 10 just says something is wrong with your pc and it gave me that blue screen with the repair button. I tried so many things I turn up the core voltage slowly and it reached to 1.4 with no sign of improvement, still doesn't boot, I changed the core/gt voltage mode Auto, Adaptive, Override, I changed the cpu loadline calibration control and cpu gt loadline calibration control from mode 1 to mode 2 and 3. I don't know what to do I'm really new to overclocking please help me.

Thank you
 
Solution
Ok. I guess I feel I should clarify the testing method further since I'm getting the feeling you don't know much about overclocking.

Set the voltage to 1.4v and multiplier to 47 to get a 4.7Ghz clock.

Run the heat test (small FFT test) in Aida64 and no other stress. Make sure only the heat test is check marked and nothing else.
Run the test for 15 minutes.

If you crash, freeze, hang, etc. before the 15 minutes is passed, you cannot get 4.7Ghz at 1.4v and it is not a good idea to go any further than 1.4v with the cooler you have.
If this happens you will have to settle for 4.6Ghz at 1.3v which you said seemed to be able to run the stress tests just fine.

However, if you do pass the 15 minute mini test stop the test and restart the PC...

QwerkyPengwen

Splendid
Ambassador
if you've tried higher voltages with no effect, then it just might be that you lost the silicon lottery and that's as far as that chip will go.
Not all CPU's are created equal and they are only guaranteed to run at the stock speed that is set for them and the turbo speed.
most everyone can at least overclock all cores to the turbo speed, but anything beyond that is not a guarantee at all and is entirely up to your luck in getting a good chip from what we like to call "The Silicon Lottery" since it's a gamble of sorts whether or not you get something good and capable.

And since the turbo of the 7700K is 4.5Ghz, and you are overclocking to it, that just goes to show you at least got an average chip and didn't get screwed with a chip that didn't overclock at all.

however, it could be that you missed something or didn't do something right, so here's a guide you can follow.
Click each of the blue links in the original post to show each section of the overclocking guide.
https://www.overclock.net/forum/5-intel-cpus/1621347-kaby-lake-overclocking-guide-statistics.html

Hope it all works out for you and good luck.
 

QwerkyPengwen

Splendid
Ambassador


doesn't sound like you're hitting any thermal walls then.

Refer to the guide I linked you to and see if there's something in there that's different than what you did or if there's something in there that you missed doing entirely. Could be why you're not getting past the turbo speed on all cores.
 
Nov 28, 2018
21
0
10


yes, but it just crashed at 4.5ghz, but i looked at the temp when it crashed and it was 60 c. I thought that 4.5ghz was stable, but i guess it wasn't since I just clocked it at that speed today.
 
Nov 28, 2018
21
0
10



I'm not really sure what i did wrong, but my overclock present right now which is 4.5ghz just crashed, but the temps were at 60 to 70 c. but the guide you linked says that I need to add more voltages. So now i'm at 4.6ghz and still benchmarking if it will crash
 
Nov 28, 2018
21
0
10


Also will I get better temps if I make the fans facing switch on the radiator. Like I turn the fan's front facing the radiator instead of the back of the fan facing the radiator?
 

QwerkyPengwen

Splendid
Ambassador


........... you lost me there.
If you're radiator is mounted to the front of the case, then the fans need to be oriented so that they front of the fan faces the front of the case and the back of the fan (where you can see the cross beams and the wiring) faces towards the inside of the case so that air is being pulled into the case from the front and being exhausted out the top/back of the case.

But those temps you are speaking of. is that when being stressed or at idle on the desktop?
 
Nov 28, 2018
21
0
10


My cpu cooler is put on the top of my case and the temps are when at 100% load idle temps are 30-50 c
 

QwerkyPengwen

Splendid
Ambassador
Try going up to 1.35v and see if you can get 4.7

Like I said earlier though, just because other people can get higher overclocks like 4.8 at lower voltages doesn't mean you will and this is all because of the silicon lottery. Sometimes you just get an average chip that overclocks to slightly lower than what other people get and at a higher voltage. It's all a part of the gamble.

Good to know that you're sitting pretty at 4.6 which means you are getting 0.1Mhz faster than turbo on all cores.

But jump straight to 1.35v and try 4.7. If it doesn't take bump it up to 1.4v

If it sticks at 1.35v, then you try going up to 4.8
If it doesn't stick 4.8 then you go up to 1.4v.
If it doesn't stick to 4.8 at 1.4v then you revert back to 1.35v at 4.7 and know now that you aren't going any further.
If 4.7 doesn't stick at 1.35v however, then go up to 1.4v.
If that doesn't stick then revert back to 4.6 at 1.3v since that was the last known stable overclock you had.

Once you've found the limit of the overclock proceed to the next phase.

The next phase is start lowering the voltage by 0.01 increments until it's unstable then put it back up one increment to where it was last stable. This is how you ensure you are getting the lowest voltage possible for your overclock.
(Example: lowering from 1.3 to 1.29 then to 1.28 etc etc.)

Once you've found the lowest stable voltage you can run at, proceed to go through the process of extensively testing the stability of the overclock using different tools and benchmarks to stress the CPU.

There is no shortcuts with this last part.
Just spend the time that is necessary and be patient.
Typically you would run a small FFT stress for up to 2 hours.
If you make it past the 2 hour mark, you can proceed to do workload stressing where you load it down with a full stress rather than just an FFT stress and let it run for anywhere between 4 to 6 hours.

if all goes well, then you're set and you've found your max stable overclock.

From here you can try to control the efficiency of the overclock by adjusting the load line calibration.

Use CPU-Z to take note of the current active voltage during stress tests.

What you want is for it not dip below the voltage that you've set for the overclock.

When the voltage drops below the set number in the BIOS, it's called V-Drooping.
This is bad and can cause system instability.

What is OK to happen is when the voltage fluctuates upward.
This is called overshooting and it simply means occasionally the voltage jumps upward for a second or two during use.

The higher you go on the load line calibration, the less V-Droop there is but the more overshooting there is as well in return.

So when adjusting the load line calibration, you want to get it to stop V-Drooping but you don't want it to go crazy with the overshooting.

So once the V-Drooping stops happening is where you want to leave the LLC.

After that, you are definitely done with your overclock.
 

QwerkyPengwen

Splendid
Ambassador
it means it booted. which can go wither which way.
Refer to what I said in my post.

When I say "if it sticks" I'm not saying "if it boots"
What I'm saying is "if the overclock is stable"

Clearly that is not stable. Go up to 1.4v and try again doing some stress tests to see if it is stable or not.
Refer to my post for basic information.
 
Nov 28, 2018
21
0
10


How long do I run a stress test for to know that it's stable?
 

QwerkyPengwen

Splendid
Ambassador
Ok. I guess I feel I should clarify the testing method further since I'm getting the feeling you don't know much about overclocking.

Set the voltage to 1.4v and multiplier to 47 to get a 4.7Ghz clock.

Run the heat test (small FFT test) in Aida64 and no other stress. Make sure only the heat test is check marked and nothing else.
Run the test for 15 minutes.

If you crash, freeze, hang, etc. before the 15 minutes is passed, you cannot get 4.7Ghz at 1.4v and it is not a good idea to go any further than 1.4v with the cooler you have.
If this happens you will have to settle for 4.6Ghz at 1.3v which you said seemed to be able to run the stress tests just fine.

However, if you do pass the 15 minute mini test stop the test and restart the PC and go into the BIOS.

Change the voltage down by one increment to 1.39v and run the heat test for 15 minute again.

If you pass the test, drop the voltage down by another increment to 1.38v and try again.

repeat this process until you experience any instability at which point you will revert the voltage back to the previous setting that passed the 15 minute test.

Once you have brought the voltage back to it's previous setting that passed the mini test, proceed to do an extended heat test that lasts for 2 straight hours.

If you experience any level of instability during the extended test bring the voltage back up again.
If this happens to be happening when running 4.7Ghz at 1.4v then this means it's not stable and you must go back down to 4.6Ghz at 1.3v

If this is happening at a lower voltage than 1.4v then as I just said, you bring it up a notch making sure you aren't going pass the 1.4v limit we have set ourselves.

once you can pass a 2 hour run of the heat test at whatever speed you had to settle for in order to do this, (whether it be 4.7Ghz at 1.XX voltage or 4.6Ghz at 1.XX voltage) proceed to tick mark the other options for CPU stress testing in Aida64 so that when you run the test, it runs the full load test.

Run this full load test for at least 6 hours straight.

If you can pass 6 hours without failure of any kind you can continue to let it go for another 2 hours just to be sure if you want.

Once you've passed both the 2 hour heat test and the 6-8 hour full load test you can continue onward with what I said in my earlier post about the load line calibration.

In fact, it might be best to configure the load line before all of this by doing short heat tests to monitor the V-Droop.

This methodology of incrementally lowering the voltage applies the same if you happen to have to settle for the 4.6Ghz at 1.3v overclock.

Doing it in reverse if 1.3v happens to not actually be stable enough to pass a 2 hour heat test where you instead bring the voltage up by one increment until you pass the test.

I hope I made this easy enough to understand for you.

If for some reason you feel you need more articulately detailed explanation of any part of this please let me know so I can elaborate further if need be so that you understand it.
 
Solution
Nov 28, 2018
21
0
10


Is the heat test in Aida64 the stress cpu? Also yes I'm very new to overclocking as I just started yesterday.

Okay I ran a stress test for 15 minutes at 4.7ghz and 1.4 voltages. I will lower the voltages and run the 2 hour benchmark tomorrow. When I find the sweet spot, I'll give you guys an update.
 

QwerkyPengwen

Splendid
Ambassador
No. You don't lower the voltage and waste time with a 2 hour benchmark.
read carefully.

I said if it passes the 15 minute test, lower the voltage by that small increment and run another 15 minute test and keep doing this until it fails.
then when it fails, bring it back up an increment THEN run a 2 hour stress to make sure that voltage is good.

If it fails a to reach 2 hours then bump it back up another increment until you pass the 2 hour heat test. Then proceed to do a full CPU stress test for 6-8 hours.

Doing a 2 hour test every incremental decrease in voltage will just result in you wasting a lot of time.