Originally Posted by jdorje
Has anyone ever done it both ways and compared?
I only put clu on the die, and i got great temp improvement. But on haswell you have to be a bit more careful about getting lm everywhere. If i remounted i might try putting on both and comparing. But placing the ihs on top of the die isn't that precise either...i imagine if you go with a water block instead you can get it more consistent.
And a real question! If you remount do you need to replace the clu? Logically it is still just liquid sitting there and you shouldn't have to...
Ive done both ways but its only going to be a fact on my system. Like ive said before there's so much inconsistencies and manufacturing tolerances that it might vary greatly. Ive done it stock, then bare die/tim then bare die/clu (temps were IDENTICAL btw but it pumped out rapidly at 1.2v.), then i did ihs/clu.
ihs/clu was 53°C and bare die was 46-47°C IN MY CASE. It really depends on the waterblock, the waterblock convex, the die flatness. Some people said it made no difference. I did it the same day, same tests, and i got way better temps bare die.
This time im going straight from stock to delid and bare die with CLU to see what temp drops i get. RIght now at 1.21v im at 59-60°C under water. Yes i know its quite high, its also because my case ambien is 30-31°C with NO AC.
Oh and no need to reapply, and on haswell and ivy, you can just nail polish over the resistors or wtv and youll be fine. A coat or two fully cured is perfect, then even if the clu runs a tiny bit its totally fine.