This baby ( http://forum.bitcoin.org/index.php?topic=7216.msg156703#msg156703
) is probably 60 times quicker than that slowstra. And it is not even the latest generation of my rigs. At the same it probably cost as much as the slowstra if you take into consideration wages of all those PhD's during many month it took them to build this slowstra thing.
Actually that slowstra is a perfect example of how some uni researchers work. Surely they had lots of fan, spent loads of money on wages for themselves, wasted load of time on collaborating with mb manufacturers, designing and manufacturing custom cases, etc... etc... etc... all this while for a fraction of the price and more importantly time they could have a few 4U off the shelf cases with 2 -3 cards each. A Nx4U system could have been built in just 2-3 days (including waiting for the components to arrive).
Idiots with PhD's... or maybe I am missing something.
Also, they didn't design it for any specific purpose. If it's for a business then the comp should serve it's primary function and no other. It's like miners that try to split a system between gaming and mining, why? Just build a miner for $600 and build a gaming comp for $1500. You'll get two computers that are about the same price as the one split-purpose system, but you'll get better results from both. The Slowstra is just a monstrosity. It can't be sold commercially, who would buy it? It isn't useful for gaming, there's no way to use all that power. It just doesn't make sense. It's no even useful for research, you could do the same job in several systems for less, and with greater reliability. Consider, if the Slowstra mobo explodes, the whole system is shot. If a cluster of computers has a mobo explode, it might lose 5% of it's total operating capacity. Nope, Slowstra is just stupid and impractical. Vladaimir's system design makes way more sense. So much so, I built my cluster of computers based on the same design.