Originally Posted by rul3s
That makes a lot of sense.
Friday when I received the noctua I've made some tests, with H115i PRO before dissasemble I got 67º max temp on realbench.
30 minutes later, with NHU12A, got 74º on max temp with same test.
At first I was thinking about thermal paste, just need time to sit down, but, heatsink was really "hot" so thermal-conductivity was working good.
Maybe PBO was reaching higher than with CLC, I dont know now, but 100% agree that test should have been made with fixed voltage/freqs, now it's to late.
But I'm happpy with the change, NHU12A is not as big as I thought and its working pretty well and silent, so it's OK, but I'll keep in my mind that H115i PRO was performing better, louder, but better.
Keep in mind CPUs under air coolers should reach max temp in a few minutes, like 2 or 3, maybe 5 minutes. That is assuming CPU goes to 100% load as soon as test starts .. and case airflow is supplying air at or near room ambient to cooler.
Liquid cooling takes longer because it take longer for coolant to reach stabled load temp. The more coolant in the loop the longer it can take.
So your 30 minutes means it to that long for coolant to reach stable temp.
Below is graph of test run on my old i7-980. You can see how fast the temp stabilizes when load starts and ends. This was on PH-TC14PE cooler w/ TY-143 fans on it running about 1200rpm. Case had 2x front and 1x bottom TY-140 fans speed controlled by CPU temp idling at 600rpm and at about 900rpm when CPU was under load and heating up.