Assist with Overclocking i7 5820k (AIDA64 evaluation)
Assist with Overclocking i7 5820k (AIDA64 evaluation)
Hello everyone, I'm just starting out with overclocking. I followed some instructions from a Linus' Tech Tips video because my Mobo and CPU are the same as his. I have an i7 5820k and a X99-A II, paired with a simple Hyper Evo 212. When he runs the AIDA64 stability test, he says if it fails before 10 minutes, I should lower the multiplier or, if my CPU temperatures stay around 80°C, I should increase the voltage slightly.
The issue is: the test passes, but my CPU temps reach nearly 100°C sometimes!
How should I handle this? Am I safe because the test passed? Or should I try lowering the multiplier as if the test didn't succeed?
Edit: video link: https://m.youtube.com/watch?v=Nj4ixZE7EWo
I'm "stuck" at 7:10.
fedpul :
Hi Fraga500, I recommend checking stability using prime95 for 24 hours to confirm the voltage range where you remain stable. If your memory isn’t OCing, consider performing in-place FFTs during a custom blend test (8-4096K, 15 minutes per FFT). Completing all FFTs in about 21 hours will take roughly an hour and a half. Success without errors after this period indicates full stability. Keep in mind that setting priority to 10 may lock your computer for the duration of the test; before proceeding, monitor temperatures during a short torture test with small FFTs at priority 1 to ensure everything is within safe limits. If temperatures are acceptable, you can then proceed with the 24-hour evaluation.
Thank you for your quick reply, rhysiam.
The current settings are 1.3 and now 1.27. Should I further reduce the voltage? The ratio is at 34 (125MHz).
Is it advisable to continue testing? Is it safe to maintain the 92°C peaks on CPU Core 3 during a 15-minute test?
What adjustments should I consider to lower these peaks?
Thanks for your prompt response, rhysiam.
The current settings were 1.3, now adjusted to 1.27. Should I further reduce the voltage? The ratio stands at 34 (125MHz).
Continuing testing is advisable? Is it safe to push the CPU core 3 to 92°C during a 15-minute test?
What adjustments should I consider to reduce these peaks?
Thanks
Updated: ratio reduced to 34 (125) - around 4.25GHz, and voltage set to 1.23. Peak temperatures reached between 80-82 during stability test. Is this excessive?
That's acceptable, though still fairly high.
If you've conducted extensive testing with AIDA64 and the system remains stable at 4.25Ghz and 1.23V, my concern would be...
Could you lower the voltage further while keeping stability intact?
If you're willing to invest more time, keep reducing the voltage until instability appears. Then increase it slightly again. If you can bring it down to or below 1.2V and stability persists, your current cooling setup should suffice, and you should typically maintain temperatures under 75°C most of the time.
A superior cooler would not only lower temperatures on your existing output but also enable higher voltages without overheating. Assuming 4.25Ghz remains stable at 1.23V (requires a thorough stress test), you're likely to reach stable performance between 4.5Ghz and 4.6Ghz at around 1.3V or higher. Enhanced cooling could support this voltage, potentially allowing a reliable 24/7 operation. However, this would probably add about 5%-8% more performance loss. So it might not be worth the effort.
Updates have been made to reduce the ratio to 34 (125) - around 4.25GHz, and the voltage to 1.23. Peak temperatures reached between 80-82 during the stability test. Is this acceptable?
It’s fine, though still on the higher side.
If you’ve conducted extensive tests with AIDA64 and the system remains stable at 4.25Ghz and 1.23V, my question is...
Can we lower the voltage even further while maintaining stability?
If you’re willing to invest more time, keep reducing the voltage until instability appears, then slightly increase it again. If you can bring it down to or below 1.2V and stability persists, your current cooling setup should suffice, and you should likely see temperatures under 75°C most of the time.
A better cooler would not only lower the temperature on your existing component but also let you operate at higher voltages without overheating. If 4.25Ghz is stable at 1.23V (you’ll need a thorough stress test), then reaching 4.5Ghz to just over 1.3V seems achievable. Improved cooling could support that voltage, enabling a reliable 24/7 operation. However, this would likely boost performance by 5%-8%, which might not be worth it.
Thanks for the advice!
I assumed the temperature readings were high, so I further reduced the voltage. I’m cautious and want to ensure safety while cooling isn’t optimal. I also live in a very hot country (Brazil), so my PC temperatures usually rise even more.
Currently, the ratio is still 34 at 125MHz (4.25GHz), but the voltage is now 1.18. What are your thoughts? Should I consider increasing the ratio while keeping this voltage?
Also, what are the downsides of lowering the voltage? The system might eventually crash if it can’t maintain the required ratio.
Hi Fraga500, please consider using prime95 to verify stability over a 24-hour period to confirm the voltage range where you remain stable. Should your memory not be OCed, perform in-place FFTs during a custom blend test (8-4096K, 15 minutes per FFT). This process will take approximately 21 hours; if you complete it without errors, you are fully stable. Keep in mind that setting priority to 10 may lock your computer for the duration of the test. Before proceeding, monitor temperatures during a short torture test with small FFTs at priority 1 to ensure everything is within safe limits. Only after confirming normal temperatures should you run the full 24-hour assessment.
Fedpul :
Hello Fraga500, I recommend checking your stability with prime95 for 24 hours to confirm the voltage at which you remain stable. Should your memory not be overclocked, consider performing in-place FFTs during a custom blend test (8-4096K), which should take about 21 hours per FFT. If you complete all tests without errors after this period, you’re fully stable. Keep in mind that setting the priority to 10 will make your computer unusable for the duration of the test. Initially, monitor temperatures during a short torture test with small FFTs at priority 1 to ensure they stay within safe limits. Only after confirming normal temperatures should you proceed with the full 24-hour evaluation.
Also remember that your Haswell CPU supports AVX 2.0, and current Prime95 versions will create significant heat on this processor. Some argue that true stability requires maintaining performance across all workloads, but we’ve observed consistently over the years that heavy AVX usage leads to excessive heat. For stress testing, I wouldn’t recommend version 28.5 or newer of Prime95 for a Haswell OC.
Others have different views, but I can tell you that pushing high temperatures with AIDA and AVX workloads will raise temperatures significantly. The trade-off is clear: lowering the voltage may improve stability but risks damage if it becomes too low. If you notice crashes or freezes, it’s a sign the voltage is insufficient for your frequency. Reducing it won’t harm components—it can actually extend their lifespan. If you’re patient and willing to experiment, pushing the voltage down further could yield a more stable result.