Unfortunately such a cluster doesn't exactly optimize sheer performance because there's lots of communications back and forth, no shared data model, etc. Granted, due to the low memory constraints, the data could be replicated on each Beowulf compute node; however, then different costs are assessed for intrathread communications. Furthermore, I'm guessing you just mean put a few processors in each box and get 10 or so of them, but at that point is a Tesla solution not better?

Is this question worth reposting so that it filters up the list for tomorrow? I don't want to break any written/unwritten rules with respect to being annoying/irritating/pesky.

This is a good point. Thank you. Again though, at these prices, it's almost worth rechecking calculations than paying for such an insane markup, no? I look at the various GTX cards, then at the Tesla, and I come away unimpressed at the differences per unit cost. So, no, I'm not sure that I don't need ECC. I'm willing to entertain the various possibilities though.