6700k boosted to 4.5 GHz, 40º idle at 87º load using AIO NZXT x 62 kraken cooler
6700k boosted to 4.5 GHz, 40º idle at 87º load using AIO NZXT x 62 kraken cooler
As the title suggests, I’m experiencing high temperatures at 4.5 ghz during overclocking, while idle runs smoothly. Using a high-performance liquid cooler like the NZXT Kraken X62 (280 mm radiator) doesn’t help much. When playing Watchdogs 2 with all settings maxed out on my GTX 1080, I’m constantly getting alerts that my CPU hits 87º in under a minute of gameplay. I’ve set the voltage to Adaptive because I’m unsure what setting works best for this first overclock. Both turbo boost and hyper threading are enabled.
I believe there are a few things I could try to fix this: switch the cooler to performance mode or reapply the thermal paste, as I changed its orientation a few times during installation for better cable management. This might have altered how the paste was spread.
Please share your thoughts before making any changes, thank you ahead of time.
Ok, so I ran the Stressed test on the CPU and the maximum voltage shown by CPU-Z was 1.472. I thought that was too high, so I stopped the stress test and switched to manual mode, where it was 1.32 to 1.35. I’m now doing the tests again, hoping for the best results.
Now I have two more questions:
1. How long should I run these stress tests? What’s the suggested time frame? Which options should I check on the Bench tab in CPU-Z? I noticed two boxes—one for single-thread and another for multi-thread—and a section listing CPU references with two buttons: “bench CPU” and “Stress CPU.” Which ones should I focus on?
2. When I lower the voltage to...
What voltage for Vcore is shown in the software you're using? It might involve a mix of thermal paste distribution and a high Vcore setting. Are there any memory overclock settings active (XMP Profile enabled)? Yes, I used Corsair Vengeance LED DDR4 at 3000MHZ with XMP enabled. Also, I'm not sure about the exact voltage since I've heard Adaptive Voltage is common—it helps avoid overheating without overloading the CPU. It varies based on load. I'd appreciate some advice on safe limits for a "safe" setting.
I tend to keep my Vcore at 1.35V but some prefer 1.4. My 6600K remains stable at 4.6G with a Vcore of 1.29V.
What is the highest Vcore reported in CPU-Z during stress tests? Adaptive voltage might be pushing too much Vcore and increasing temperatures. I suggest using manual Vcore to determine the voltage needed for stability at 4.5G. Set it at 1.35V and check if you remain stable with temperatures under 80°C. If stable, reduce it slightly to 1.34V and verify again. Continue lowering until you find the minimum voltage that keeps you stable. Once identified, return to adaptive mode and adjust the BIOS offset so it doesn’t apply more Vcore than necessary.
Also - when XMP is enabled, it increased my temperatures by about 7°C. On my Gigabyte board, there’s a "Relax OC" option for memory that helped bring my temps back down.
Possible reasons for your high temperatures include insufficient thermal interface material, overly high adaptive Vcore, or XMP activation. I recommend starting with manual Vcore testing to find the lowest stable setting.
I tend to cap my Vcore at 1.35V, though some prefer 1.4. My 6600K remains stable at 4.6G with a Vcore of 1.29V. What is the highest Vcore shown in CPU-Z during stress tests? Adaptive voltage might be pushing too much Vcore and increasing temperatures. I suggest using manual Vcore to determine the stable voltage at 4.5G. Set it to 1.35V and check stability with temperatures under 80. If stable, reduce to 1.34V and verify. Continue lowering until you find the minimum needed for stability. Once identified, adjust adaptive settings in BIOS and apply an offset to avoid overvoltage. Also, note that XMP enabled boosts temps by about 7°C; using "Relax OC" on memory helped lower them. Possible causes for high temps include insufficient thermal interface material, excessive Vcore in adaptive mode, or XMP activation. I recommend starting with manual Vcore tuning to find the optimal setting. Thanks for the helpful and clear advice—it’s great to have such guidance as a beginner! I’ll make sure to follow up after I’m back home and keep you posted.
Here’s a revised version of your text with the same length and structure:
Ikloz confirmed! Your detailed response is perfect for someone just starting out, and I’ll make sure to follow all the adjustments when I get home. I’ll keep you updated on my progress.
Yesterday I drafted a quick summary of my overclocking process. Although there are various approaches to reach the same result, this method works well for me. Pay close attention to your temperatures since they’re currently a concern for you.
First, install CPU-Z to verify BIOS settings and monitor Vcore. Next, use HWMonitor to track core temperatures. Then run OCCT for stability and maximum temperature checks. While overclocking, disable turbo, EIST, and all energy-saving modes (c-states).
Set the Vcore to 1.35V. Adjust your multiplier to 42 and perform OCCT-Small Blocks for 30 minutes, checking temperatures. If they stay below 80°C, increase the multiplier to 43 and repeat. Continue this process until temperatures exceed 80°C (or you encounter a blue screen or error in OCCT)—then reduce the multiplier by one.
While testing, also monitor Vcore in CPU-Z. If it drops to 0.01V or lower under your BIOS setting, you’re experiencing Vdroop and should apply some LLC. Aim for enough LLC to limit Vdroop without causing Vboost (voltage spikes).
Once stable, run OCCT-Large for 3 hours to confirm stability. If you encounter a blue screen or core errors, lower the multiplier by one and retry. Keep repeating until you find a stable multiplier where OCCT-Large runs smoothly for 3 hours.
After reaching a stable setting with temperatures under 80°C, I’ll gradually reduce Vcore by 0.01V at a time and re-test OCCT-Large. Continue this until stable operation is achieved while keeping temperatures below 80°C and Vcore low.
At this stage, you can fine-tune your Vcore further. Lower it incrementally and test OCCT-Large again until you reach a reliable configuration. Once that’s done, you may re-enable EIST and possibly C-1E c-state. Now that you know the required Vcore, you can use adaptive voltage control and apply any necessary offsets to achieve stability.
I prefer keeping my Vcore at 1.35V but some users are okay with 1.4. My 6600K remains stable at 4.6G with a Vcore of 1.29V. What is the highest Vcore reported by CPU-Z during stress tests? It seems adaptive voltage might be giving too much Vcore and increasing temperatures. I suggest using manual Vcore to determine the voltage needed for stability at 4.5G. Set it at 1.35V and check if stability is maintained with temperatures under 80 during stress testing. If stable, reduce it to 1.34V and verify again. Continue lowering until you find the minimum voltage required for stability. Once identified, return to adaptive mode and adjust CPU-Z's maximum Vcore accordingly. Also, when XMP is enabled, it increased my temps by about 7°C. On my Gigabyte board, there’s a "Relax OC" setting on memory that helped bring temperatures down.
Potential reasons for your high temperatures include insufficient thermal interface material, excessive Vcore in adaptive mode, or XMP activation. I recommend starting with manual Vcore testing to find the lowest stable voltage.
Regarding your stress tests: I ran them for [insert duration] and observed a max reported voltage of 1.472. Since this is quite high, I stopped the test and switched to manual mode, testing between 1.32 and 1.35. I’m now proceeding with stress tests again, hoping for better results.
For your questions:
1. How long should I run the stress tests? What’s a typical recommended duration? Which CPU-Z tab sections should I examine—especially the bench CPU and Stress CPU options?
2. Once you reach stable performance at 4.5G, how do I cap the adaptive voltage? Is this similar to setting an offset voltage? How can I calculate the appropriate offset value?
Here is the rewritten version:
I ran the stress test on the CPU and the maximum voltage reported by CPU-Z was 1.472, which seems quite high. I decided to stop the test and switched to manual mode, where it stabilized between 1.32 and 1.35. Now I’m conducting the tests again, hoping for better results.
I have a couple more questions:
1. How long should I continue the stress tests? What is the suggested time frame? Which settings on the Bench tab in CPU-Z should I focus on? I noticed two categories—one for single-threaded and another for multi-threaded performance, plus references to various CPUs.
2. Once the voltage drops to around 4.5 GHz and becomes stable, how can I control the maximum adaptive voltage? Is this similar to setting an offset voltage? If yes, how do I find the correct offset value?
I prefer 1.32V to 1.35V instead of 1.472V. A lower Vcore might help with stability.
Also, I don’t rely on CPU-Z for stress or stability checks. I use CPU-Z mainly to verify BIOS settings and monitor the Vcore. For stress testing, I prefer OCCT. I usually run OCCT-Small for 30 minutes to ensure temperatures stay below 80°C, then proceed with OCCT-Large for 3 hours to confirm stability. Opinions vary on what defines a stable test and which program works best. OCCT is my go-to tool.
Lastly, I’m not very familiar with adaptive voltage settings in the Asus BIOS since I use a Gigabyte motherboard. It seems you might have set the CPU Core Voltage to adaptive, then changed the offset mode to "-" and entered the desired offset value into the "CPU Core Voltage Offset" field.