I believe push is better because of the way fan blades are designed to create forward pressure...I feel like they would be less efficient at working against backward pressure. Try it and tell us
, this is one of those questions that's been around forever...it makes <2c difference from what I've seen from conflicting tests...so whichever is easier is really best.
Testing done!
Constants: 3600mhz 1.38-1.39v, CPU/NB 2600mhz 1.21v, Prime runs of around 15 minutes, room temp about 75c constant. Far from scientific, but it will do.
Pull on top, Push on bottom.
The Pull Config...
The Push Config...
Well holy cow. Push (in this config at least) is good for about 2c! I call this a good test too, as you can see from the HWmonitor my TMPIN2 (mobo) temp remains a constant 39c (and yes, it does change if the temp does). By the way, the voltage hops around from 1.38-1.39...it's 1.39 99% of the time, except for the 1% when I took my screenshot, lol.
So Mort' I owe you an apology, and to the OP, I was wrong, push is better. I can only imagine ECO decided on the pull config for convenience of installation. Let it be know, not all cases will fit the rad to the case as I have without some sort of spacers or modification. Not only that, I had to root around for some screws to properly mount it like this. Standard case fan screws will mount it as shipped. So, a strong push/pull with matched fans is still best, but how do you control them variably without a fan speed controller. Next up is the push like I have done. And finally leave it as it was shipped.
For the record people trying to compare these results to the H50, I'm not using the factory ECO fan, I'm using a beast of a Delta fan controlled (turned down) by the BIOS speedfan. Thats why it was about 100 RPMs faster with the hotter result, it varies based on temps. As for this test, it was cooler running with less RPMs as set by my BIOS based on the temps.
A 2c win for Push!