Odd Heat Issue on Ivy-Bridge
Odd Heat Issue on Ivy-Bridge
Adjust the position so the tubes are at the bottom, if possible. This could be the main problem—airlock. Sometimes these coolers aren’t fully filled with coolant and an air pocket forms. Even though the pump is running, it might just be spinning the same coolant without moving it. I removed the radiator while Prime95 was active but couldn’t change temperatures no matter how I oriented it. The highest point was always on the block, outlet inlet at the top, tubes at the top, and nothing affected the temps. Any suggestions? Since this cooler is bigger than the standard 1155 model, I might install an older 775 Intel cooler and test if that helps (my motherboard supports 775 heatsink mounts). Would that be a reasonable idea?
Did you attempt this orientation? The goal is for the air pockets to exit the pump unit or block, traveling to the radiator and staying at the top of the tank. Your I/O ports and tubing should prevent them from pulling the air pocket back through the loop. The radiator must be positioned higher than the pump, with the tubing facing downward as shown.
Did you consider this orientation? You want the air pockets to exit the pump unit or block and flow into the radiator, staying at the top of the tank. Your I/O ports and tubing shouldn’t pull the air pocket back through the loop. The radiator should be higher than the pump, with tubing facing downward as shown. I didn’t set it up that way, but I held it upside down as much as possible (the bottom of the rad was the highest point) and temperatures stayed unaffected.
I'm struggling with some ideas right now. Did you consider going back to the 212 if you return to the expected normal temperatures, or are you also concerned about cooling issues? How big was the temperature spike when it was adjusted? Was there any damage from being delidded that might explain what happened?
I'm running low on ideas right now. I also wonder if going back to the 212 would help if you're expecting normal temperatures or if cooling problems persist. How much of an impact did the bump in temperature have when the 212 was moved? I can rotate it a bit while it runs, but that's not something I suggest doing. Was there any damage from being dropped, especially since it was on a concrete floor without padding?
The 212 ended up badly damaged after falling off my desk onto a hard surface. It looks like it got a solid hit to the bottom, causing it to bend severely. That’s why I considered the stock 775 cooler—it’s the next best option. The drop likely caused similar issues. The lid was installed without any damage, and it was done long before I acquired this model or faced this problem. I only noticed the temperature difference after the bump, but I don’t have proof that moving the 212 caused it.
I plan to visit my local Micro Center after work to get another 212 Evo and some thermal paste. It could be my i5 that’s failing, maybe a bad batch of paste, or possibly the H80i GT is malfunctioning. Hopefully, swapping the cooler and paste separately will help identify the issue. Alternatively, if nothing improves, I might have to try increasing voltage and clock speed until the temperatures cause the CPU to fail before upgrading.
I acquired another 212 Evo along with some Notcua NT-H1 paste. Without altering any settings, I’m experiencing low to mid 50s temperatures.
Now it’s time to reinstall the watercooling using the new paste (without removing the CPU or reapplying under the IHS) and observe the results.
UPDATE:
It seems the issue stemmed from increasing the 212 Evo, which likely created an air gap between the IHS and heat sink, followed by applying the Coollaboratory Liquid Copper paste after a fresh application of the problematic paste. I reinstalled the H80iGT with the Noctua NT-H1 paste, and after 10 minutes of running Prime95 26.6 smallFFT, my peak temperatures were between 48-54-55-53. Now everything is working well!
That's definitely a fascinating result, and it's great to see you've figured it out.
I find it particularly intriguing since the Coollaboratory Liquid Copper is claimed to be the top paste for heat transfer, not the typical "liquid metal" type. I'm not sure if this was a bad batch or if it just struggles on the die, but I'm glad I identified the issue.
I revisited the overclock settings and managed to reach 4.5GHz with random voltages, achieving a max temperature of 68°C after more than 8 hours of prime95 small FFT.
Later this evening I'll experiment with PLL, agent, and IMC voltages to see if stability improves. At 4.2 there were noticeable drops in FPS and frame time due to CPU usage in Rise of the Tomb Raider, but going up to 4.5 mostly resolved that problem.
At 4.2 I experienced fluctuations in performance depending on how much CPU was being used during Rise of the Tomb Raider, whereas 4.5 has largely eliminated those issues.
I'm really aiming for stability at 4.6, as that could be the final tweak needed to make it run perfectly!