Question All overclocking attempts unstable

Eden 1337

Distinguished
Jan 13, 2017
11
0
18,510
Hello, I recently upgraded my GPU to a GTX 2070 but I'm CPU bottlenecked in a lot of games so trying to squeeze out some extra performance from my ageing hardware!

CPU: Intel i5 7600k
CPU cooler: Corsair h100i v2 AIO
Motherboard: Asus STRIX Z270F Gaming
Ram: Corsair 16gb LPX 3000mhz (XMP enabled)
SSD/HDD: WD Black SN850x 2tb
GPU: Nvidia 2070 Rog Strix 8GB
PSU: 650W Corsair VS
Chassis: Corsair Carbide 400C Windowed
OS: Windows 10 64 bit
Monitor: LG 27GL850-B 1440p

I've tried a few combinations in BIOS such as:

42x, voltage 1.250 (OCCT fail)
45x, voltage 1.275 (OCCT fail)
45x, voltage 1.250 (OCCT fail)
48x, voltage 1.250 (BSOD, no boot)
etc

but every time it fails the OCCT stress test and the CPU also seems to max out at 100% very quickly with basic tasks. The computer only behaves normally when I set the CPU settings to auto. Activating the built in 'Asus Optimal' preset also causes system instability.

My CPU + cooler should be able to handle a small overclock, but it seems I'm missing something? Any help is appreciated.
 
Last edited:
I recently upgraded my GPU to a GTX 2070 but I'm CPU bottlenecked in a lot of games so trying to squeeze out some extra performance from my old hardware!
Intel i5 7600k
Corsair h100i v2 AIO
Asus Z270-F
Corsair 16gb LPX 3000mhz (XMP enabled)

When posting a thread of troubleshooting nature, it's customary to include your full system's specs. Please list the specs to your build like so:
CPU:
CPU cooler:
Motherboard:
Ram:
SSD/HDD:
GPU:
PSU:
Chassis:
OS:
Monitor:
include the age of the PSU apart from it's make and model. BIOS version for your motherboard at this moment of time.

Speaking of upgrades, did you run DDU to remove all GPU drivers(intel, AMD and Nvidia) in Safe Mode then manually installing the latest GU driver sourced from Nvidia's support site in an elevated command, i.e, Right click installer>Run as Administrator?
 
42x, voltage 1.250 (OCCT fail)
Id expect to hit 42x at 1.25V, but perhaps you should to start a bit lower and work your way up slowly from 38x. The 7600K should turbo boost one core up to 4200MHz with no overclock. I haven't had any problems running earlier CPUs (i7-4770K, i5-4670K, G3258) up to their turbo boost frequencies all-core with 1.25V. N.B. I tend to stick at 1.25V Vcore max for long life.

As an experiment, switch off XMP and revert to stock JEDEC RAM speed. With no adjustment to Vcore, set an all core multiplier of 39x and see what happens. The BIOS should raise Vcore automatically and I'd expect the system to be stable.

Next, raise the all core multiplier to 40x (again no manual Vcore setting) and still with XMP switched off. Check Vcore in AIDA64, HWMonitor, etc. It may well be pushing up past 1.20V towards 1.25V. Increase the multiplier again to 41x, then 42x, but do not exceed 1.30V automatic Vcore. Check for stabiity each time.

Most BIOS are very conservative and apply far too much automatic Vcore as you raise the all core multiplier. Although some people run their 7600K up to 1.40V @ 48x or 50x, I'd stick with 1.30V max for the time being.

If your system is stable at 1.25V manual Vcore with a 42x multiplier and no XMP, you've established a base point to continue. The system might be stable at 2133 or 2400MT/s JEDEC RAM speed, but not at XMP 3000MT/s. Which do you value more? CPU speed or memory speed.

You might need to change the Uncore frequency (if that option is available) to improve memory stability when raising the CPU multiplier. There's often a sweet spot you need to establish with Uncore, but I've never bothered tweaking it. See what other people have achieved in the past, but remember they may have been using "golden samples", de-lidded CPUs or run extremely high voltages.
 
  • Like
Reactions: Eden 1337
Id expect to hit 42x at 1.25V, but perhaps you should to start a bit lower and work your way up slowly from 38x. The 7600K should turbo boost one core up to 4200MHz with no overclock. I haven't had any problems running earlier CPUs (i7-4770K, i5-4670K, G3258) up to their turbo boost frequencies all-core with 1.25V. N.B. I tend to stick at 1.25V Vcore max for long life.

As an experiment, switch off XMP and revert to stock JEDEC RAM speed. With no adjustment to Vcore, set an all core multiplier of 39x and see what happens. The BIOS should raise Vcore automatically and I'd expect the system to be stable.

Next, raise the all core multiplier to 40x (again no manual Vcore setting) and still with XMP switched off. Check Vcore in AIDA64, HWMonitor, etc. It may well be pushing up past 1.20V towards 1.25V. Increase the multiplier again to 41x, then 42x, but do not exceed 1.30V automatic Vcore. Check for stabiity each time.

Most BIOS are very conservative and apply far too much automatic Vcore as you raise the all core multiplier. Although some people run their 7600K up to 1.40V @ 48x or 50x, I'd stick with 1.30V max for the time being.

If your system is stable at 1.25V manual Vcore with a 42x multiplier and no XMP, you've established a base point to continue. The system might be stable at 2133 or 2400MT/s JEDEC RAM speed, but not at XMP 3000MT/s. Which do you value more? CPU speed or memory speed.

You might need to change the Uncore frequency (if that option is available) to improve memory stability when raising the CPU multiplier. There's often a sweet spot you need to establish with Uncore, but I've never bothered tweaking it. See what other people have achieved in the past, but remember they may have been using "golden samples", de-lidded CPUs or run extremely high voltages.

Hi, I've followed those instructions! I bumped it up to x42 and now hwmonitor is showing:

CPU VCORE max 1.296 V
CPU VID max 1.322 V

It appears stable but under load in OCCT the temps go up to around 85c. Is that okay? Where do I go from here? I was hoping to achieve 4.5ghz but it seems I'm already at the 1.3v threshold. Are these current settings faster than the stock x38 with 4.2 boost?
 
I ran 1.416 volts for a few years with my 7700k @ 5Ghz. But I also delidded it (just thermal paste though) to get one of the cores to match the others, and semi-custom water cooling.

For a 24/7 4.8Ghz I would say you need closer to 1.35, which is what the VID is calling for, 1.322 (That is the requested voltage)

OCCT temps at 85C are pretty good, how about when gaming?

A big factor in stability is often load line calibration. Which prevents voltage droop, but can also drive the CPU voltage higher than expected when load suddenly drops. Try adding a little.
 
CPU VCORE max 1.296 V
OK, so that proves it will run a very minor overclock of 42x all core, when the BIOS automatically increases Vcore to 1.296V.

Next you have to find the manual Vcore option in the BIOS (disable automatic Vcore) and start lowering it until the system becomes unstable.

If you have the patience, reduce manual Vcore in steps of 0.010V. Try 1,290V, then 1.280, 1.270, and so on, until the system becomes unstable.

I wouldn't be surprised if you could get close to 1.200V manual Vcore at 42x without XMP (maybe even lower) before you get problems, but a lot depends on the quality of your CPU.

Alternatively, set a manual Vcore of 1.300V (disable automatic Vcore) and increase the multiplier to 43x, then 44x, and so on until the system is no longer stable.

If 45x (or 44x) is unstable at 1.300V, increase manual Vcore to 1.310V, 1.320V, etc., but stop at 1.350V.

It's up to you to decide how high to take Vcore. The higher you go, the shorter the CPU lifespan. At 1.400V, the CPU might last 3 years (complete guess). At 1.300V it might last 4 years. At 1.25V 5 years.

Then again, the CPU might suffer electro-migration damage after only 18 months at 1.4V. At that point you'll have to reduce or disable the overclock and maybe go back to stock voltage. See what other people say.