Overclocking i7-6700k, need some assistance!
Overclocking i7-6700k, need some assistance!
Hi, I just assembled a new system. It’s an Asus Hero Maximus viii with an i7-6700k processor, 16GB DDR4 RAM at 3200 MHz. I’ve configured it to run at 4.6 GHz and the voltage is set to 1.28. I’m seeking assistance for two issues. First, when monitoring via software like HW Monitor, the voltage fluctuates instead of staying steady, sometimes exceeding 1.3 volts, which concerns me because it could be unsafe. Second, I struggle to maintain a consistent 4.6 GHz during gameplay; it often drops to 4.0 GHz and then jumps back to 4.6 GHz. I tried using manual settings without XMP or auto, but I’m unsure if there’s anything in the BIOS I can adjust to keep it stable at 4.6 GHz. My previous machine was an i7-4790k and it worked well, but I don’t remember the exact steps. The BIOS is similar but not identical. Any advice would be appreciated.
The voltage variation is typical. When AVX instructions or other complex CPU operations are involved, Vcore will increase by about 0.13 volts, sometimes more. This explains why it's advised against running CPU stress tests with synthetic data, as they can overstress the processor excessively. An application-based test that mimics real-world usage offers a more accurate simulation, providing a steady load rather than sudden spikes. Use RoG Real Bench to achieve this realistic environment.
The voltage variation is typical. When AVX or other complex CPU operations are involved, Vcore may increase by about 0.13 volts, and possibly more. This explains why it's better not to perform CPU stress tests using synthetic data, as they can overwork the processor excessively. An application-based test that mimics real conditions offers a more accurate simulation, even though the voltage increase remains limited. RoG Real Bench is recommended for such realistic scenarios.
2. Games rarely put the CPU under heavy strain. Even GPU stress tests like Fumark only reach 40% CPU usage. RoG Real Bench remains an ideal choice in these cases.
3. Running continuously at 4.6 GHz isn't wise. It's similar to driving a car at high speed in a school zone—no benefit and excessive wear on the system. If you still wish to try, consider disabling Intel Speed Step and C States through a web search.
4.. A direct jump from stock settings (like .4.2 turbo) to 4.6 GHz isn't advised. It's safer to make gradual changes so you can observe trends in settings, temperatures, and voltages as the multiplier increases.
Keep monitoring with HWiNFO; it’s a comprehensive tool for tracking performance.
RoG Real Bench is highly effective for stress testing and evaluating stability in multitasking environments.
Links:
https://www.hwinfo.com/
http://dlcdnmkt.asus.com/rog/RealBench_v2.44.zip
The voltage variation is typical. When advanced CPU instructions are used—such as AVX or similar—the Vcore will increase by about 0.13 volts, sometimes more. This explains why it's better not to run CPU stress tests with synthetic data, as they can overstress the processor for extended periods. An application-based test that mimics real usage provides a more accurate workload. Games usually don’t put the CPU under heavy load either; even GPU stress tests like Fumark only reach about 40% CPU usage. RoG Real Bench is ideal for realistic scenarios.
Running constantly at 4.6 GHz is not recommended. It’s similar to driving an A6B 6500 at high speed in a school zone—no benefit and excessive wear on the system. If you still wish to try, consider disabling Intel Speed Step and C States through a web search.
It isn’t advised to jump directly from stock settings like .4.2 turbo to 4.6 GHz. It’s safer to make gradual changes so you can observe how settings and temperatures/voltages change as the multiplier increases.
For ongoing monitoring, HWiNFO is excellent, and Real Bench lets you stress-test the CPU while testing stability in a multitasking environment.
Links:
https://www.hwinfo.com/
http://dlcdnmkt.asus.com/rog/RealBench_v2.44.zip
It seems acceptable but I really wouldn't want to be completely certain at this stage. Comparisons vary with different monitoring tools and stress tests. You might find yourself being synthetic stable yet failing with applications, particularly when multitasking. If your tests don't cover AVX and I'm unsure if the specific AIDA test you used does, then you might not be observing the maximum voltage.
I operate at 1.38 and it rises to 1.41 without AVX... but when AVX is present, I see values between 1.49 and 1.50, though they last only milliseconds. I'm not very confident about voltage until I run a Real Bench benchmark (the third and fourth components include AVX).
I feel more secure with voltage when I can complete two hours of multitasking without failing, as previous OCs that failed under RB showed. It's hard to match the value knowing a CPU can run stable synthetically unless the aim is to rank on web leaderboards. Usually, the goal is stability and comfortable temperatures while running apps. RB confirms I'm stable enough, allowing higher clock speeds compared to synthetic results, which often cause overheating.
It seems acceptable, though I wouldn't be completely sure at this stage. Results vary with different monitoring tools and stress tests. You might find yourself being synthetic stable yet failing when using applications, particularly under multitasking conditions. If your tests don't cover AVX and the specific AIDA test you used isn't included, you may not be observing the full voltage range.
I operate at 1.38 and it rises to 1.41 without AVX... but with AVX I see values between 1.49 and 1.50 for short durations. However, these remain brief, lasting only milliseconds. I feel more confident about voltage stability only after completing a Real Bench test (the third and fourth components include AVX).
I’m not fully satisfied with multitasking reliability until I can run it continuously for two hours without failure, as previous models that reached 24 synthetic stable units failed under RB. It’s hard to accept the value when knowing a CPU can perform well in tests unless you aim to rank on websites. My main goal is stability and comfortable temperatures while running applications.
So I followed the programs you recommended. I tested at 4.6ghz with 16GB RAM, ran for 15 minutes and passed. I plan to extend the duration to confirm stability. The voltage reading is low—set it to 1.28 in BIOS, but HWmonitor shows it’s lower than expected. This is reassuring because a low VID is better than a high stable processor running above 4.6ghz. I’m uncertain about pushing this chip further, but I’d rather avoid that. I prefer low temperatures and voltage with moderate performance boosts. I’ll check if stability is possible at 1.275v. Thanks for your support and the two programs. This was a challenging journey—I’ve been diving deep into overlocks before, but nothing like this chip. It’s natural to acknowledge unknowns, which is why I started this thread.