Textured pin heatsink:

Textured pins have emerged as one of the two main calling cards of Swiftech coolers. The first being those nice base finishes. I got a big response last time by saying the bases "weren't polished", here is a line by line explanation right from the man, about how they're actually finished.

"Our bases are hand polished on a granite table with Diamond-based polishing
paper. Here is a step by step manufacturing process (true for all our coolers):

A) CNC flycut. The surface is 100% flat, to 1/1000", but the CNC leaves machining marks (circular traces as seen in most coolers). Flycutting leaves the surface very "bright" which sometimes get people mixed up with a mirror finish. In fact, if you measure the surface with a surface roughness gage, you will usually obtain between 30 and 40 MSF (micro surface finish). This rough finish seriously affects thermal performance. Example: I gained 11Mhz overclocking with the Alpha 8045 after lapping the factory finish.

B) The HS is then lapped on our 48" Lapmaster machine. This is a 48" cast iron rotating table, lubricated with lapping compound. We machine lap from 36 to 48 heatsinks at a time, in about 15 to 20 minutes. This process removes all the machining marks, and leaves the surface perfectly smooth but with a "mat surface finish".

C) Finally, the heatsinks are hand finished on granite stones, covered with diamond polishing paper, and lightly lubricated with solvent (Naphtha). This process gives the "shine" to the surface. Depending how long you hand polish the product, you get down to a 1 MSF (perfect mirror). It takes 30-45 seconds per HS to get to a 6-8 (our current MSF), and it may take up to 3 minutes, and using additional lubricants, to get to 1 MSF. The final polishing (whether it's 8 or 1) is really cosmetic. Performance wise, current CPU's do not see the difference."

Well there you have it, I stand corrected, Swiftech heatsink bases are polished - just not to a mirror shine. ;-)

What about Dust?

There is a little voice in the back of my head that is kind of wondering what's going to happen with all those threaded pins over time. I mean sure, they do increase surface area and turbulence, and hence performance, but what happens when dust starts settling down? I've seen some pretty nasty heatsinks after as little as two months of use that were caked with dust, and all their surfaces were smooth. Somehow I'm wondering just how big of a dust trap the textured pins will be over time. Dust as we all know, will impact on the performance of any heatsink if enough of it builds up. With all those little nooks (per pin), here's hoping that the textured pin technology doesn't act like a super dust filter.

Swiftech has improved the width of the retention mechanism from the previous version but I am still a little curious about how well the force is being directed downward. Since the clips are centered on the heatsink, the force the heatsink exerts is also centered. Ideally, the force being exerted by the heatsink should be off-center, and over the location of the processor core.

The bottom line is that the MCXC370 takes a bit of extra time to set properly, else it may look like it is mounted properly (and it will work just fine), but you may be missing out on several degrees of extra cooling as the heatsinks' forces are concentrated on one edge of the die. A good analogy would be shifting the weight of your body to one foot while standing. Both of your feet are still touching the ground, but one foot is taking more of your bodies weight.

These issues (while not in any way life threatening) cropped up on our test system, where we could adjust the spring tensioning to produce the best thermal interface insofar as forces over the core were concerned.