ChromeTusk :
Just my personal experience, but I have noticed a slight change on a few PCs I have upgraded in the past. When I changed a down-draft cooler to a tower-style cooler, the "system" or "northbridge" temperature usually rose 1-4 degrees C. There are plenty of heat sinks on today's motherboards that could use the extra airflow. Just make sure you exhaust all that hot air instead of re-circulating it.
Indeed, the case and airflow layout probably makes a larger difference than just the orientation of the CPU cooler fan.
RonKinNJ :
I have to question the results of the Noctua NH-L12. Perhaps you had a defective sample, or maybe a problem with thermal contact. FrostyTech tested it on AMD and Intel platforms with 125 watt and 85/150 watt loads, respectively. Given the form factor, it fared well.
At least I thought so, and bought one. Compared to the inadequate cooler that came with my box, it is doing very well. i7-2700K @ 3.5. Even under load, I have yet to see full speed on the fans.
That's not a fair comparison. Your Sandy Bridge runs cooler than the Phenom used in this test, especially since it isn't overclocked, so your experience will be better than the reviewer's. However, Patrick does say in the review that a less ambitious thermal envelope, such as a HTPC or stock CPU, would have adequate cooling from the Noctua.
Also, Frosty uses a CPU heat simulator, so how applicable that is to real-world is debatable. Their heat pad could have a much flatter and finer finish than a retail chip is likely to have ( for better thermal conduction, ) it might dissipate heat in a different pattern across the surface, etc.