I was just about to post a link to Martin's Lab. He has shown that if one or the other go with push. As far as shrouds he has data showing that they are almost worthless and then he has data showing they make a small effect. Just depends on the exact scenario: what radiator, what fans, and wether you are rocking push or pull. Now 2 degrees doesn't sound like a whole lot but when you start improving 2c here and 2c there and then 1c over there it starts stacking up. For example I have custom small fan over my VRM heatsink. When activated that will drop the CPU temp nearly 6c. Then supposedly the new Apogee HD that I just got cools 2c better than the previous design.
So read and review and do some tests. Just an FYI when you are not setting up a lab grade environment you need to increase your margin of error. Maybe a 10% or more margin. That is why I rely heavily on lab quality reviews by people like Martin. Another person did a 360 rad round up and that is how I came to decide I would hunt down a TFC X-Changer instead of buying a XSPC RX360.
So read and review and do some tests. Just an FYI when you are not setting up a lab grade environment you need to increase your margin of error. Maybe a 10% or more margin. That is why I rely heavily on lab quality reviews by people like Martin. Another person did a 360 rad round up and that is how I came to decide I would hunt down a TFC X-Changer instead of buying a XSPC RX360.