F5F Stay Refreshed Power Users Overclocking Stress test stable, unstable during gaming - 3570k. Need suggestions.

Stress test stable, unstable during gaming - 3570k. Need suggestions.

Stress test stable, unstable during gaming - 3570k. Need suggestions.

H
HotMilkTea
Member
204
08-08-2025, 08:37 AM
#1
I recently acquired a new GPU, and it became clear that the main issue in my setup wasn't the previous GPU (an R9 290), but rather my CPU. This is particularly noticeable in newer titles such as ACOd. Given that I don’t have much room for further upgrades soon, I chose to attempt overclocking my older 3570k.

My goal now is to reach a clock speed of around 4.4Ghz, which is already partially achieved as mentioned in this post. However, gaming still triggers warning messages (no BSOD, just event ID 20 warnings). For testing, I used OCCT (medium dataset), realbench, and prime(small FFT), all of which passed without issues for at least 8 hours each time, at least twice. I also ran MemTest to confirm stability.

In everyday use and some MATLAB tasks, OCCT works smoothly—no errors, no crashes, no stutters. But when playing games (currently testing ACO), I still encounter some warnings. Below are the settings I maintained:

- XMP is active for my 8GB DDR3 memory, delivering the rated speed of 1600MHz (2x4GB Corsair Vengeance). I prefer keeping it enabled since altering memory timings and voltage feels risky. I also retained C states and EIST functions.
- For OCCT, current settings are:
- Main: XMP enabled for 8GB DDR3, 1600MHz
- Multi: x44
- V core: currently at 1.285, jumps to 1.32 under full load
- Turbo mode: disabled
- Current temperatures under normal load: 70°C–74°C
- Prime temps: 85°C–88°C
- Idle temps: 35°C–37°C

I haven’t made any other adjustments at the moment, and I’m not sure I’ve overlooked anything. My search for the ideal vcore began at 1.25 with everything else unchanged. At that voltage, I was unstable during stress tests. I increased to 1.26, which improved things temporarily; Prime still failed on OCCT and realbench. Then I went to 1.27, passing both tests, but Prime still crashed after 6 hours on OCCT. After that, I tried 1.275, which passed OCCT and all other benchmarks. I then began testing games—at 1.270, two warnings appeared within an hour of gameplay (expected since OCCT also failed). At 1.275, I remained stable for nearly four days with around 10 hours of playtime. Eventually, a game crash occurred; I reviewed the logs and found a new warning. Eventually, I settled on 1.285, which is now stable but hasn’t been tested extensively.

In conclusion, I’m seeking advice from someone more experienced about this OC setting. Am I on the right track, or am there another option I’ve missed? I’ve lost some confidence because I feel I’m getting close, and every hiccup makes me doubt.

Thanks for reading my lengthy explanation!

Edit: My post’s vcore values were incorrect due to a typo—I forgot to prepend 1.x before typing. Apologies, and I updated the post with the correct settings.
H
HotMilkTea
08-08-2025, 08:37 AM #1

I recently acquired a new GPU, and it became clear that the main issue in my setup wasn't the previous GPU (an R9 290), but rather my CPU. This is particularly noticeable in newer titles such as ACOd. Given that I don’t have much room for further upgrades soon, I chose to attempt overclocking my older 3570k.

My goal now is to reach a clock speed of around 4.4Ghz, which is already partially achieved as mentioned in this post. However, gaming still triggers warning messages (no BSOD, just event ID 20 warnings). For testing, I used OCCT (medium dataset), realbench, and prime(small FFT), all of which passed without issues for at least 8 hours each time, at least twice. I also ran MemTest to confirm stability.

In everyday use and some MATLAB tasks, OCCT works smoothly—no errors, no crashes, no stutters. But when playing games (currently testing ACO), I still encounter some warnings. Below are the settings I maintained:

- XMP is active for my 8GB DDR3 memory, delivering the rated speed of 1600MHz (2x4GB Corsair Vengeance). I prefer keeping it enabled since altering memory timings and voltage feels risky. I also retained C states and EIST functions.
- For OCCT, current settings are:
- Main: XMP enabled for 8GB DDR3, 1600MHz
- Multi: x44
- V core: currently at 1.285, jumps to 1.32 under full load
- Turbo mode: disabled
- Current temperatures under normal load: 70°C–74°C
- Prime temps: 85°C–88°C
- Idle temps: 35°C–37°C

I haven’t made any other adjustments at the moment, and I’m not sure I’ve overlooked anything. My search for the ideal vcore began at 1.25 with everything else unchanged. At that voltage, I was unstable during stress tests. I increased to 1.26, which improved things temporarily; Prime still failed on OCCT and realbench. Then I went to 1.27, passing both tests, but Prime still crashed after 6 hours on OCCT. After that, I tried 1.275, which passed OCCT and all other benchmarks. I then began testing games—at 1.270, two warnings appeared within an hour of gameplay (expected since OCCT also failed). At 1.275, I remained stable for nearly four days with around 10 hours of playtime. Eventually, a game crash occurred; I reviewed the logs and found a new warning. Eventually, I settled on 1.285, which is now stable but hasn’t been tested extensively.

In conclusion, I’m seeking advice from someone more experienced about this OC setting. Am I on the right track, or am there another option I’ve missed? I’ve lost some confidence because I feel I’m getting close, and every hiccup makes me doubt.

Thanks for reading my lengthy explanation!

Edit: My post’s vcore values were incorrect due to a typo—I forgot to prepend 1.x before typing. Apologies, and I updated the post with the correct settings.

R
Reltzy
Member
111
08-08-2025, 08:37 AM
#2
The CPU voltage is very high... it's extremely high!
R
Reltzy
08-08-2025, 08:37 AM #2

The CPU voltage is very high... it's extremely high!

I
iTzKohaPvP_Ez
Junior Member
35
08-08-2025, 08:37 AM
#3
Wait wait wait... you run that Chip with a Vcore of more than 2,00V? That must be a readout Error because usually your Chip runs at 0.8 to 1.4V and trust me if I say a Chip with 1.4V is getting really damn hot. Under normal room temperature a chip instantly dies if you hit it with 2.2V or more...
I
iTzKohaPvP_Ez
08-08-2025, 08:37 AM #3

Wait wait wait... you run that Chip with a Vcore of more than 2,00V? That must be a readout Error because usually your Chip runs at 0.8 to 1.4V and trust me if I say a Chip with 1.4V is getting really damn hot. Under normal room temperature a chip instantly dies if you hit it with 2.2V or more...

R
Reepety
Senior Member
374
08-08-2025, 08:37 AM
#4
Yeah, sorry about that mistake. I really meant 1.x instead of 2.x. I'll fix the voltages.
R
Reepety
08-08-2025, 08:37 AM #4

Yeah, sorry about that mistake. I really meant 1.x instead of 2.x. I'll fix the voltages.