Fractal R5 Window, NZXT Kraken X61 280mm AIO. Stock 140mm Fractal fans, all fans 600-900rpm.
I initially ran the aio on top /push, with just the 2x fans as intakes, did all kinds of cpu/gpu stress tests.
Then switched to front mount /push, then changed to front /pull. There's a difference.
With the top mount I got 70 cpu/ 78 gpu.
With front push I got 70 cpu /80+ gpu (capped and throttled).
With front pull I got 67 cpu / 74 gpu.
I attribute this to several reasons. First, with lower than @ 1300rpm fans, pull actually benefits due to the broader area of rad used, push generally has a motor mount blockage, seen easily by the 'dust donut' if you don't clean it often. So I was actually getting better effective airflow and resultant temp drop.
Also, with push, the fans are mounted in front of the rad, which has 2 affects. First is it disperses most of the static pressure as it travels through the rad, you still get the cfm, but it doesn't reach the back of the case or the gpu very well. The second being there's a door in front, so there's little direct room in front of the fans.
By moving the fans behind, I get the second reason. Static pressure. There's now nothing between the fan and gpu, the fans are much closer to the gpu, so the gpu benefits like a bandit from increased airflow. Cpu also benefited from the fan placement as there's much more room behind the door allowing for better vaccum properties, better airflow
Front pull optimizes gpu temps, front push doesn't, top can benefit the cpu in push with high rpm fans or benefit the cpu in pull with low rpm fans, does little to nothing for the gpu as that's at the mercy of the intake fans either way.