Going to one HD 4547

Peter T. Breuer Actually, with the Intel-supplied heat sinks for the Xeon processors, the nominal heat sink is a copper plate about 2.5" x 3.5" x 7mm thick with a 32 aluminum fins parallel to the 3.5" edge. The fins are about 2" high and 3.5" long. I.e., about 448 square inches of surface area (each fin has two sides). This is a lot more than the mere 8.75 square inches of the copper plate, so while in principle the plate helps cool directly, it is only about 2% of the cooling area.

The airflow clearly does not cool the copper plate directly, at least not very much, but cools the fins that conduct the heat from the copper slab.

Intel strongly suggest blowing the cool air into the fins from the front of the tower, confining the air to go through the fins and not around them by using their "processor wind tunnel" (PWT) which is a duct that completely surrounds the heat sink including the output of the cooling fan, and at the rear where the heated air escapes. Thus all the air from the fan cools the fins of the heat sink, and does not escape around the sides of the heat sink buttembly.

Now you could pop a hole in the top of the PWT and blow down or up there, but if you do, than about half the air would go out the front, and half the air would go out the back. The air going out the front would be opposing the general airflow through the tower that is primarily from front to back. So the flow through half the heat sink fins would be less than designed. Also that air would cause a loop of hot air from the processor fan intake through the front half of the heat sink fins and out the front of the PWT only to be sucked into the cooling fan again (diluted somewhat, I admit), which is not such a great idea. It is leaving to chance something that is better handled by design.

-- .~. Jean-David Beyer Registered Linux User 85642. V PGP-Key: 9A2FC99A Registered Machine 241939. ^^-^^ 14:55:00 up 71 days, 8:50, 3 users, load average: 4.29, 4.25, 4.26