Originally Posted by Taeric

Deadlines are not the issue as they're set so far in the future that most any system can make them. However, the calculations progress from one generation to the next as soon as the prior generation finishes, so the sooner that happens, the faster to overall project progresses. Slowing down WUs to squeeze a few extra points (30%ish) out of a HT CPU causes each of your WUs to get submitted later than the would if you did not run the second instance, and that has the potential to slow down the overall research despite cranking out more points.

This info comes straight from Stanford, so it's not a personal crusade of mine. I'm just passing along the best way of doing things in their opinion, which is something I highly value since they're the ones who put this whole thing together.

Within out team competition, HT is definitely a bad idea because it counts as 2 CPUs but only produces about 30% more points than running one instance that counts as 1 CPU. The above issues aside, you team will get the most points if they can fold with 8 physical CPU cores.