My 8700k Overclock results - 4.9ghz/1.24v - room for improvement?

jayleonis

Honorable
Nov 18, 2018
108
0
10,590
New to overclocking but have been researching and wanted some general thoughts on my results

asus Z370-a
gtx 1071ti
16gb g.skill trident z
i7-8700k
750w psu

No adaptive offset
AVX offset - 2
LLC - level 5
4.9ghz over clock
1.24 vcore

Ran prime 95 for 10 min and real bench for 10 min back to back. These were my min max values

Vcore - min - 1.84 max 1.232
CPU - min 33° - Max - 82°

During the test my Vcore stayed at 1.20. Does this mean I can lower my voltage even further because to run under load all it needs is 1.2?


 
Solution
That is not even a BEGINNING of what you need to do to test the viability of that overclock.

Here is the quick and dirty version:

Quick and dirty overview of overclocking validation procedure.

Set CPU multiplier and voltage at desired settings in BIOS. Do not use presets or automatic utilities. These will overcompensate on core and other voltages. It is much better to configure most core settings manually, and leave anything left over on auto until a later point in time if wish to come back and tweak settings such as cache (Uncore) frequency, System agent voltage, VCCIO (Internal memory controller) and memory speeds or timings (RAM) AFTER the CPU overclock is fully stable.

Save bios settings (As a new BIOS profile if your...
That is not even a BEGINNING of what you need to do to test the viability of that overclock.

Here is the quick and dirty version:

Quick and dirty overview of overclocking validation procedure.

Set CPU multiplier and voltage at desired settings in BIOS. Do not use presets or automatic utilities. These will overcompensate on core and other voltages. It is much better to configure most core settings manually, and leave anything left over on auto until a later point in time if wish to come back and tweak settings such as cache (Uncore) frequency, System agent voltage, VCCIO (Internal memory controller) and memory speeds or timings (RAM) AFTER the CPU overclock is fully stable.

Save bios settings (As a new BIOS profile if your bios supports multiple profiles) and exit bios.

Boot into the Windows desktop environment. Download and install Prime95 version 26.6.

Download and install either HWinfo or CoreTemp.

Open HWinfo and run "Sensors only" or open CoreTemp.

Run Prime95 (ONLY version 26.6) and choose the "Small FFT test option". Run this for 15 minutes while monitoring your core/package temperatures to verify that you do not exceed the thermal specifications of your CPU.

(This should be considered to be 80°C for most generations of Intel processor and for current Ryzen CPUs. For older AMD FX and Phenom series, you should use a thermal monitor that has options for "Distance to TJmax" and you want to NOT see distance to TJmax drop below 10°C distance to TJmax. Anything that is MORE than 10°C distance to TJmax is within the allowed thermal envelope.)

If your CPU passes the thermal compliance test, move on to stability.

Download and install Realbench. Run Realbench and choose the Stress test option. Choose a value from the available memory (RAM) options that is equal to approximately half of your installed memory capacity. If you have 16GB, choose 8GB. If you have 8GB, choose 4GB, etc. Click start and allow the stability test to run for 8 hours. Do not plan to use the system for ANYTHING else while it is running. It will run realistic AVX and handbrake workloads and if it passes 8 hours of testing it is probably about as stable as you can reasonably expect.

If you wish to check stability further you can run 12-24 hours of Prime95 Blend mode or Small FFT.

You do not need to simultaneously run HWinfo or CoreTemp while running Realbench as you should have already performed the thermal compliance test PLUS Realbench will show current CPU temperatures while it is running.

If you run the additional stability test using Prime95 Blend/Small FFT modes for 12-24 hours, you will WANT to also run HWinfo alongside it. Monitor HWinfo periodically to verify that no cores/threads are showing less than 100% usage. If it is, then that worker has errored out and the test should be stopped.

If you find there are errors on ANY of the stability tests including Realbench or Prime95, or any other stress testing utility, you need to make a change in the bios. This could be either dropping the multiplier to a lower factor or increasing the voltage while leaving the multiplier the same. If you change voltage or multiplier at ANY time, you need to start over again at the beginning and verify thermal compliance again.

A more in depth but general guide that is still intended for beginners or those who have had a small amount of experience overclocking can be found here:


*CPU overclocking guide for beginners


 
Solution
That is not a stability test.
Heat soak can take up to an hour to equalize. Your case gets hotter the longer you operate it at high load.
So running a few minutes will not give an real temp or stability for hours of gaming. Your case and cpu will get hotter with the video card producing heat also.
The above stated test are good.
I go one further.
OCCT. It can stress CPU/GPU or both.
You can also set shut down temps, in case it gets too hot during test.
Run at least 24 hours if your computer is used for anything but gaming.
I usually stress for 3-4 days , as my computers are used for folding 24/7/365 @100% load.
https://www.majorgeeks.com/files/details/occt.html

 

jayleonis

Honorable
Nov 18, 2018
108
0
10,590


okay so just did the prime95 26.6 thermal test for 15 min. Showing a Max CPU Temp of 82°C. Ill run the Realbench now for several hours and see. Although in real life, id never be maxing my cpu out for 3+ hours so im not sure how relevant it would be for me.
 

jayleonis

Honorable
Nov 18, 2018
108
0
10,590



What do you mean use for "folding"? what is that? Im wondering if those tests are super relevant to me since Id never really max out my cpu for an extended period of time like that.
 
4.9GHz at 1.24v is very good....I can only manage 1.278v for a 4.9GHz on my 8700K...But do test for longer on Prime95 version 26.6 both blend and small FFT's as that will shake out any issues with the overclock.

Overall I would also run small 20 to 30 minute runs on AIDA and Realbench just to make sure all is okay across a variety of tests...but overall, if you can keep stable at 1.24v, that is good and you can start to try and drop that further but it should also allow for a better and higher overclock at lower voltages...

Best of luck on the overclocking journey..
 
Even if you never max out your CPU, even if you only use 20% of your CPU, you need a STABLE overclock.

Some decide a "stable" overclock is one that boots to windows and allows them to run any application without any crashes or restarts. The issue is micro errors that can happen as a result of an unstable overclock. I have experienced this first hand with my 1st overclock of my 7700k.

Micro errors are when your OS hit small errors that do not reach the level of crashing a game or your OS. You never know they happened, until its too late. After months pf use, they will build up and slowly corrupt application and OS files. At some point in time, your applications or OS will begin to exhibit problems. My initial signs were that I couldn't connect to online servers,for 2 of the games i played. You may think, no big deal. The problem is uninstalling and reinstalling the application doesn't solve the issue, you must wipe and do a fresh windows install. This could be a small issue to a large one depending on,many factors. If you don't mind reinstalling windows and everything on your system every 6-12 months then don't worry about stability testing. Otherwise, follow the guidance Darkbreeze gave you.
 
Folding is scientific calculations for protein folding simulations.
https://foldingathome.org/
These are very demanding on your hardware so everything must be 100% stable.

An unstable overclock can cause random errors, corrupt your hard drive, corrupt windows or give errors or incorrect results in every day tasks.
Make sure your overclock is stable.

Everyone want to brag about their high overclocks(me too mentality), but they are useless ,unless you are 100% stable.
I've been overclocking for 22 years and held a few records along the way, but stability is the main concern.
 

jayleonis

Honorable
Nov 18, 2018
108
0
10,590



Ah I had no idea about that. Im guessing if the overclock and survive 8 hours of realbench/prime without crashing then there would be no micro errors present? or are there other ways to look for it such as noticing 1 core not operating at 100% during the stress test
 

jayleonis

Honorable
Nov 18, 2018
108
0
10,590
So I am running real bench now. And it’s mid way through and says at the bottom “instability detected” and says “result hash match” in the left window. I tried to click show results but says I
Must finish a full run at least once. Should I finish to see results or cancel now that I know it satana instability detected
 

jayleonis

Honorable
Nov 18, 2018
108
0
10,590
So I stopped the test, upped the voltage from 1.24 to 1.28. During the 15 min thermal test temps hit 88. Is this too high? I didn’t move onto the real bench because I wasn’t sure

Moved back down a little to 1.25 and lowered frequency to 4.8. Temps hit a high of 82. And have passed real bench for 2 hours now. I’d like to go back to 4.9 but the temps scare me. Althoigh if its only under max load hitting 88 maybe that’ not so bad since it would rarely hit that in daily use
 
Yes, it is too high. I would HIGHLY recommend that you stop, reset your bios to the default settings, CHECK to see if there is a NEWER bios version available and upgrade to it if there is. Then, start over with your overclock and move incrementally up until you find the plateau at which you cannot remain remain both thermally compliant

(NO higher than 80°C. It is still "safe" up to 85°C, as far as you are not immediately damaging your CPU, but you WILL incur damage to the CPU at 85°C if used continuously or even now and then repeatedly, at that temperature. If you cannot remain below 80°C then the frequency is too high, the voltage is too high or the case or CPU cooling is not sufficient for the overclock.)

and maintain stability with NO errors in Realbench or any other stress utility. 4.9Ghz is what I'd consider an extreme overclock considering it is about 600mhz above the Intel specified all core turbo ( 1.2Ghz above the base clock and 200mhz above even the single core max turbo speed) and more like 800mhz above the all core turbo speed for any other Coffee lake CPU. Considering this is a 6/12 processor, not a 4/8 like previous generation i7's, you really can't expect to see the same sort of frequency gains this time around since you're multiplying things by an additional 2 cores (4 additional logical cores).

You're going to see things heat up much faster when you increase by 500mhz on a 6/12 CPU than when you increase by 500mhz on a 4/8 CPU.

I'd read the guide I outlined earlier, and the one below, and then work your way up. Being impatient is NOT the way to achieve a reliable, stable, responsible overclock that isn't likely to seriously reduce the longevity of your processor or motherboard.

https://www.tweaktown.com/guides/8481/coffee-lake-overclocking-guide/index.html
 

jayleonis

Honorable
Nov 18, 2018
108
0
10,590


so im following your advice, temps too high and got instability error, so i lowered freq to 4.8 and voltage to 1.215. ran real bench over night last night for 8 hours and it showed "instability detected" I checked hwinfo64 and it says I got 3 WHEA errors saying "cpu cache l0 errors". does that error mean anything specific like more voltage or lower freq, or just a generic instability and i have to go figure it out?
 
If there are errors it can only really mean a few things on a system that was normal/stable BEFORE the overclock (Which is why I always recommend running these tests on the system in its stock configuration first, so that you KNOW if it was stable then or not, otherwise you don't know if the instability is due to the overclock configuration or your hardware.).

Assuming the system was stable before the overclock, it generally means that either the frequency is too high, not supportable by the sample (CPU) you have, power delivery on the motherboard is not adequate (Weak VRMs) or the voltage is too low.

For you, I'm seeing the voltage you have and thinking the voltage is too low. Obviously, raising voltage tends to increase heat which might make you have to lower the overclock, which is exactly why beginners think they have some kind of golden sample compared to those most avid overclockers have, because they don't bother testing the stability or doing the thermal testing correctly. If they did, they'd realize they need more voltage to be stable at their current configuration, and therefore they'd often exceed the thermal specification and have to lower the multiplier to reach a point where it is both thermally compliant AND stable.

4.8Ghz might not be achievable for you. Or it might. Raise voltage slightly, retest. Rinse, repeat.
 

jayleonis

Honorable
Nov 18, 2018
108
0
10,590


Thats funny you say that as I read all over people bragging about heir sub 1.25v 5ghz stable OC. seems extremely unlikely. probably dont do sufficient stress testing. most probably like me, read a few quick things on how to do it and run prime for 20 minutes without crashing haha.

in regards to thermals though, is it really a big deal to go up to 85-90 during a max load stress test? Dont the chips auto throttle down to prevent damage? and especially if outside the stress tests using the setup for its main purposes, lets say gaming, if the temps never go above 70. could you make a judgement call and allow those temps for the stress test? even deb8auer says in his 8700k guide that you can go up to 95 degrees (he caveats it might affect stability)
 
Most people will never see the temps you see in stress testing.
Most software is not coded to keep the processor 100% utilized.
But some will.
The higher the temp the more chance you have of an error,and the more power needed to keep it stable.
The chances of an error increase with speed and temp.
Running at stock speed 95c is near max temp but the probabilities of an error is very very slim. Intel does rigorous testing and binning to ensure this.
Once you start overclocking, all of that is gone and you have to dial it in yourself.
Some chips are great overclockers, others not, it is all in the silicone lottery and you tinker until you get it just right. Or leave it at stock.
 
Actually, temp has nothing whatsoever to do with stability EXCEPT possibly if you are at or near Tjmax. Then things will likely become unstable OR the system will throttle.

If the system throttles, it's not going to be unstable because frequency and voltage are likely to be severely reduced.

What WILL cause instability is a poor memory configuration, a CPU clock frequency that is too high for the amount of voltage that it is set to or a voltage setting that is so high that the VRM/MC/power delivery cannot handle it.

Also, potentially, a faulty CPU, faulty memory or faulty motherboard.

I've never seen a CPU become unstable because it was hot, unless it was VERY, VERY hot. Like, over 100°C. That does not mean however that it is not suffering damage. Instability and thermal damage are entirely different things.

I would reduce that overclock to 4.7Ghz, which is the SINGLE core max boost speed, and give it about 1.3v, and test it. If it's still unstable, up the voltage by .005v or thereabouts. If it is stable, then increase the clock frequency to 4.8Ghz and test again. If it is unstable, increase the voltage and test again.

Be sure to test thermal compliance by running Prime95 version 26.6 on Small FFT EVERY time you make a change to the CPU configuration.

Before you do ANY of that though, I still do not see that you have offered us the details regarding what CPU cooler you are running or what the exact model number of your power supply is?
 

jayleonis

Honorable
Nov 18, 2018
108
0
10,590


oops i didnt see the post asking about my cooler. Im using a Kraken x62.

So i went from 1.215v to 1.23v and ran realbench again last night for 8 hours and passed. no instability errors and no errors in WHEA from hwinfo. Max temp overnight was 79 degrees. so i think im just going to leave it here and not tinker with it unless you see a reason to? In fact I may run that second prime95 test you were recommending after just to confirm stability. But I think upping voltage much more and going for 4.9 would result in 80+ degree temps.

I really appreciate your help thanks a lot! let me ask you one more thing if I could. Now that I have a stable configuration, should i start to dial in an appropriate offset? What is your opinion on the debate between using offset and just running straight at your vcore voltage 24/7. some people say it affects the longevity of your cpu, some say if its under 1.5 it doesnt matter..

I was actually running with speedstep disabled and powerplan settings on turbo mode (max cpu usage 24/7) until i recently read another one of your posts about how you should definitely not be doing that given that the ramp time from low cpu usage to max is almost instantaneous and there is no way to notice a difference. and instead saving on temps is more valuable.
 
Yes, that sounds ok, however, do you have XMP enabled for your memory and are they currently running at their advertised speed and in dual channel?

If you have not yet enabled XMP for your memory, now would be a good time to do that, and then run 4 passes of Memtest86 to ensure that the memory configuration is also stable. Memory configurations are much more likely to create microerrors than a CPU overclock, within reason of course. Memory errors are also no joke. They will quickly corrupt your data and file system. If you've passed Realbench, your memory configuration is PROBABLY fine, but it can't hurt to be sure by running Memtest.


*Resolving memory problems and setting up XMP/DOCP/AMP profiles
 

jayleonis

Honorable
Nov 18, 2018
108
0
10,590



yes i did have xmp enabled prior to all that. just ran 1 pass of memtest with no errors. ill get to the other 3 soon. noticed in another thread youve been in this space since the 80s i believe you had said. pretty impressive. really appreciate sharing all the advice!


I think i'm pretty capped at 4.8 here given thermal issues and going over 80 degress at much higher voltage. so the next thing i was thinking is creating an offset. do you find running the cpu 24/7 at an overclocked voltage shows any evidence of cpu degradation even assuming proper temps and stability. lots of people like to dial in an offset or adaptive voltage to prevent the constant 1.25 or whatever voltage even when idle.
 
No, you can't do it like that. You can't do one pass, then two later, then another at some point. All passes are not the same. The passes get incrementally harder to pass, because the memory heats up which is part of what you want to see is whether or not it can still pass all four passes even when it's heated up from the combination of mathematical and row hammer tests. Plus, the difficulty increases, from what I'm told. I have no way to verify that since I am not an electronics engineer but what I can say for certain is that I've seen plenty of memory configurations pass one to three passes and then fail on the fourth one.

So you want to run all four passes in a row, no break in between passes. Run it overnight or while you're gone to work. There's no chance of harm from you not being present. You'll come back to see it either passed or there were errors. If there are errors, bump the DRAM voltage up by about .005-.020v in the bios settings, save settings and test again.
 

jayleonis

Honorable
Nov 18, 2018
108
0
10,590


ah yes i did run a full test. it does 4 passes. thought you meant from the beginning run it 4 times lol.