Adjusting performance beyond limits using a tailored water cooling system
Adjusting performance beyond limits using a tailored water cooling system
I’m having trouble finding suitable benchmarks to gauge my performance. My readings are higher than desired, and I’m experiencing temperature spikes above 80°C despite the water loop maintaining a stable temperature. Perhaps a fresh thermal paste application would help. Could anyone provide some advice?
Using Asus AI 5 way tuner.
The system defaults to 1.375V, 4.4GHz with 100bclk. This configuration isn’t reliable for gaming. I revert to 1.35V, 4.2GHz with 102bclk, which causes my wireless connection to weaken and requires a restart to restore stability...the temperatures are too high? I then return to 1.35V, 4.2GHz overclock at 100bclk and run prime95. Temperatures remain below 78°C, though core 3 occasionally spikes to 85°C before dropping again...this is unusual since core 3 typically runs about 15°C hotter than core 6. Could a new thermal paste fix this issue?
I spent four hours experimenting with overclocking last night, but the outcomes were disappointing. The water temperature never exceeded 36°C, and the VRM stayed between 55-65°C. I aim to keep the CPU’s hottest core under 78°C and would prefer lowering the voltage if possible. The Asus 5 way optimizer pushes settings too high, leading to excessive heat and instability.
I’ll continue researching but ultimately need a reliable overclock for everyday use. I’d like to play games for extended periods without temperature concerns. My Nvidia 1080 is also water-cooled, performing smoothly at 60fps with gsync on my new Asus pg27aq monitor—though I had to overclock the GPU to +160 and memory to +200 to maintain that frame rate.
Not very acquainted with your monitoring tool, to be honest. The package temperature might actually reflect the socket temperature, which isn't directly cooled. The CPU core temperatures appear to be within normal ranges for the chip. Despite having a good watercooling system, I still notice peaks in the 70s.
It's a noticeable variation between cores. The cooler could be slightly misaligned. Also, inspect the chip closely—it might show high or low spots.
When removing the cooler, pay special attention to the thermal compound application.
I think 6th gen HEDT chips perform less well with overclocking than their Haswell-E 5th gen predecessors, so your current setup might be pushing limits. Don't hesitate to experiment with different voltages and frequencies; sometimes the bigger jumps aren't stable, but the intermediate settings are more reliable.
After reviewing more details, I figured out whether adjusting the CPU core voltage is typical for a water-cooled setup or if it's too low. Once I switched to adaptive mode with +, Auto, and set the voltage to 1.200, I configured all cores at 4.2ghz. The thermal control tool suggested dropping to 36x after reaching 75°C. Intel XTU ratings at 4.2mhz hitting 70°C max were 1742, which aligns with what you'd expect from a custom water loop. The voltage remained stable around 1.2970 volts whether idle or under load. When I used Asus Real Bench and ran a stress test, the CPU temperatures were 62°C, package 80°C, and individual cores ranging from 54°C to 75°C. This large gap between package and core temps raised my concerns about the cooling performance. The benchmark score of 150248 also matched expectations. Overall, the numbers seem consistent with a well-configured water-cooled system.
Not very accustomed to your monitoring system, to be honest. The package temperature might actually reflect the socket temperature, which isn't directly cooled. The CPU core temperatures appear normal according to the chip, and despite having a good watercooling setup, I still notice spikes in the 70s.
Thank you for your assistance Eximo! I removed the water block and re-applied the paste just to ensure safety. The temperatures haven't improved much, but they don't rise as rapidly, which seems positive.
I also observed a 5°C rise in temperatures when switching XMP on and off for my memory. I hadn't considered this setting before, but it appears cooler at 1.2v versus 1.35v. I'm now testing it to potentially lower the temps.
Right now, the Intel XTU benchmark is still around 1742 with a max of 73°C.
I also got a score of 155565 from Asus' benchmark, and the highest temperature reached about 65°C.
After the paste job, each core's temperatures remain within about 10°C of one another, which is an improvement over earlier.
It seems my VID remains fixed at 1.298v when Turbo is on, preventing me from lowering the voltage below that level in Adaptive mode. This proved beneficial since my system restarts and crashes were linked to insufficient Vcore voltage. Now I'm stable at 4.3ghz with 1.310v for over 24 hours. I've played several hours of Asus RealBench and "The Division" without issues. I'm considering lowering the voltage back to around 1.305v or less to find the lowest stable setting, though it's clear that 1.298v is too low. I've also left my Vcache voltage set to Auto since Adaptive settings haven't worked well for me so far. I haven't adjusted the cache multiplier beyond a factor of 32 and found stability with Auto mode. My main uncertainty remains about the power dissipation capacity of the EK X360 waterblock kit—how much heat can it handle and what voltage should I safely use? Anyone else own an X360 EKWB setup? If so, what voltages are you using and what temperatures do you observe?
It's a big number given my CPU specs and GPU, but I probably didn't need the other se240 radiator.
My main issue isn't just heat output, but keeping CPU temps stable under load. At 1.310v it works fine on a 6-core 6850k, but going above 1.325v causes temperatures to spike to 80C quickly. It seems more about how much heat I can extract from the CPU than total dissipation.