Push/pull on Raditators?

http://martinsliquidlab.org/2012/01/15/radiator-shroud-testing-v2/

If I remember, push and pull will out preform push by 1-2 degrees and pull by more.

Personally, I use the RX360 and RX120 with 4 GT AP15 in a push at 1000-1200 RPM (can't hear them even with the panel off) for my overclocked 2600k and GTX 580 SLI and my temps are great. Idles around high 20's to low 30s, cpu loads around 60's and gpus load around mid 40's.

All your questions can be answered (with actual testing results) by spending a few hours with Google.
Have a look at Skinnee Labs for example.

P/P > pull > push (generally), use fan spacers.
You need to know CFMs.
You need to know fan pressure.
More CFM or pressure, generally, equals more noise.
3 db = twice as loud.

Hope that gets you started.

Generally, I think most consider push/pull > push > pull in terms of efficiency. But you don't gain that much by adding a second set of fans, at a considerable cost in terms of expense and space. You probably won't have room in a prodigy case for push/pull

And fan spacers probably aren't necessary, but can be radiator specific. I've quit using them.

It's gonna depend a lot on the rad itself. On a rad with tight fin spacing 20ish fins per inch push pull is a must imo. But a rad with 7-8 fins per inch your not gonna see much improvement because the air with flow through the rad much more easily. From my experience.

Push + Pull > Push > Pull, temperature benefits will vary depending on the radiator and fan, and you might see up to a degree or two difference (each step, left to right).

It should be noted that apart from fane that are OEM'd by Nidec for servers...99% of noise specs printed on fan boxes are completely imaginary made up numbers. Nidec's specs are more trustworthy, as if they lie on the spec sheet like most others then server blades will melt.

The article on the fan tests is interesting. I had a 5-6° drop when I installed the shrouds while the CPU is under load. Its just a crappy H60 using Aerocool Sharks in Push/Pull with shrouds and a filter to keep dust out of the rad. The H60 fins are kinda tight and the shrouds on that made a big difference. I would also like to see a shroud test using the thicker/240/340 rads. Lots of dead space on a 340 rad.

I think that the benefits rise in relation to the fin density and/or the fan speed/static pressure. Powerful 38mm thick fans, for example Delta 3800rpm 144cfm/11mmHg will not really see much of a difference if you add a second set for pull, while some 120x25mm 1000rpm fans likely will show significant gains. With a radiator that has 7-8fpi and is equipped with decent fans (70cfm/2.5mmHg) the benefits will be small, but with a 20-30fpi radiator even using 110cfm/4mmHg fans, additional fans added in pull will be beneficial.
The difference can range from less than 1 percent to as much as a 20-25 percent gain, depending on the variables involved.

The Gentle Typhoon AP15s are well-loved for their excellent performance to noise ratio, and while they are certainly good performance fans, they are not the best performance fans by a long shot, they simply give an excellent mix of airflow vs volume.
The XSPC RX series radiators are generally recognized as being one of the top "quiet" radiators, hitting their best performance efficiency at around 1200rpm. While you will get better temps with higher speeds, the scale is hardly linear and gains do drop off fairly quickly.

If you want a radiator that scales well at higher speeds while still performing well at 1k-1.2krpm, I would recommend looking at either the XSPC EX series or the Alphacool NeXXos UT60/XT45 series of radiators. With either of these you can get great performance at low noise levels, but the more you crank up the fans the better cooling you will get. The same goes for push-pull in that you essentially will be able to get the same performance as somewhat higher speed push-/pull-only fans at lower speeds, and at max fan speed you will find a noticeable benefit from the push-pull setup.

Another big benefit of push-pull is that it helps minimize the fan hubs "dead spots" by forcing the air through from two different directions, with the push fans blowing some air against the fins covered by the pull fan's hub, and the pull fans are able to help better form the airflow so that the push fan's hub has a smaller dead-zone.
Push-Pull also creates more turbulence which in turn reduces the boundary layer of air against the fins, meaning more cool air contacts the metal and thus cools better as opposed to the boundary layer of warm air transferring the heat to the cool air, which is exponentially less efficient.

That test was done with a low end QP rad. You can bet the gains will go up with a rad that can dissipate more watts.