i7 6700k OC Advice
i7 6700k OC Advice
Could someone assist me in understanding these results? I'm using an i7 6700k with a Gigabyte Z170x-Gaming 6 motherboard and a Kraken X61 cooler. When adjusting the CPU core ratio to 45 (4.5GHz) and setting vcore to 1.34v, the system appears stable with maximum temperatures near 80°C during stress tests using x264, Intel XTU, Cinebench, and Handbrake encodes. Idle temps are about 38°C when Performance Mode is active on the cooler.
I've noticed many cases where users achieve higher frequencies at lower voltages and temperatures. Am I getting a faulty chip, or is my cooler underperforming? My GTX 1070 also idles around 48°C. Something feels slightly off with my performance.
Additional details that might be relevant:
- vcore LLC is set to 'Auto'
- XMP is enabled for a memory frequency of 3000MHz
- I'm in Phoenix during the middle of summer, so ambient temps are around 26°C
Occasionally using two monitors simultaneously may lead to incorrect measurements for certain MBs. You might run the other tests in AIDA64 for DIMMs and GPU, increasing the test duration to about 30 minutes. Readings should only come from AIDA64 without HWMonitor. Additionally, consider using Realbench v2.43 as an alternative benchmarker. If you succeed with that test and benchmark, your system appears stable.
Hi nromonoski
Can I really get stable performance with Vcore at 1.25V? Every tiny voltage rise leads to a big jump in heat. While not all Intel CPUs behave the same, lowering the core voltage to around 1.34V might be better. Stress tests can expose weaknesses in voltage tolerance (Vdroop). Load Line Calibration LLC is useful during heavy use but keep an eye on temps.
I rely on HWMonitor and AIDA64 trial versions for testing and monitoring.
Your Kracken X61 seems capable, but if your TIM is failing, replacement might be necessary.
Ideal temperatures are 10-15°C above ambient when idle and 60-65°C under load.
Also, ensure you have a reliable PSU—stable, efficient power is crucial.
MeanMachine41 :
Hello there,
I'm trying to figure out if it's possible to maintain stability with a Vcore of 1.25V. Every slight rise in voltage seems to cause a big jump in temperature. While not all Intel CPUs behave the same, I've found that lowering the core voltage to around 1.34V is better.
Stress testing might expose weaknesses in the voltage range (Vdroop). The Load Line Calibration LLC can be useful during high load, but keep an eye on your temps.
I currently use HWMonitor and AIDA64 trial versions for testing and monitoring. My Kracken X61 appears to be a solid machine; however, if your TIM is failing, replacement might be necessary.
Ideal temperatures are 10-15°C above ambient during idle and 60-65°C under load. Also, ensure you have a stable, high-quality PSU—efficient power supply is crucial.
Thanks for your help, MeanMachine! For your PSU concern, I own an EVGA 850W G2.
Update on the overclocking... I reduced the score to 1.25V and enabled LLC on high as you advised. The system started up but crashed during stress tests. I then increased the Vcore to 1.26V (still LLC high), and stability improved. All listed stress tests passed, with only minor temperature spikes when monitors were active.
The readings now show idle temps around 36°C and a peak of 77°C under load. I do notice occasional CPU usage and temperature spikes when the monitors are on, using tools like NZXT CAM and CPUID HWMonitor.
Does this seem better? I'm relieved the Vcore drop worked, but is it acceptable to keep LLC on high? The HWMonitor shows a consistent range of 1.248V–1.260V, while VID fluctuated between 1.232V and 1.432V. If 1.34V was too high, the spikes at 1.432V are concerning. Also, I expected lower temps with such a big voltage reduction.
You have a decent PSU and it shouldn't cause issues.
It seems you've hit the maximum of your OC range with that chip. More testing is needed.
Lower the frequency slightly to 4.4GHz at 1.25V, keep the high voltage mode active, and rerun the test. This will help check stability and performance during gaming at a lower frequency.
For the test, use HWMonitor and AIDA64 trial versions side by side on your desktop.
Run a ten-minute CPU and FPU test while monitoring all system voltages with HWMonitor.
Capture a screenshot at the 10-minute mark and submit it to IMIGUR.
Stop the test if temperatures reach 80°C.
I'm looking forward to the results.
Can you confirm if the CPU and FPU need to be checked together or individually?
I tested CPU and FPU together at the same time. The FPU reported similar temperatures by itself, so I thought it was fine to run both at once. Here are some screenshots from a 4.4GHz/1.25vcore/LLC High setup around 10 minutes into the test and one when idle (also using the original tests; results were similar, max 79°C). I’m not sure why all the voltages gave such consistent temperatures. The 4.5GHz/1.34v version reached about 80°C under load too.
Your temperatures remain within acceptable ranges at 4.4GHz with a Vcore of 1.248V.
Examine the voltages across all rails during testing. The measurements show: 3V at 2.040v, 5V at 3.367V and 12V at 0.048V.
The results suggest a significant issue with your PSU, as all rails exceed acceptable limits.
If these readings are accurate, it seems the system should have failed the test, which would indicate major instability.
Consider testing the unit under load or replacing it with a verified working unit of equivalent power to confirm.
Typically, EVGA gold-certified units are reliable PSUs and include a ten-year warranty.
Wow, that's concerning. It seems testing with a voltmeter is the most reliable method instead of using software. Could you share a screenshot for comparison? I don't have another PSU available and want to assess the variation accurately.
Oh wow! That's not good. Thanks for spotting it! It seems using a voltmeter would be the most reliable method instead of relying on software. Could you share a screenshot of your setup for comparison? I don't have another PSU available, so I'm unsure about the variations during use. Also, avoid using a multimeter unless the system is under load. My readings from HWMonitor suggest swapping the unit might be best.