On 09/16/13 09:16 PM, Jason Smith wrote:
> Thanks to Jeff Sarlo at the University of Houston... we were able to
> acquire eight 700MHz SiCortex blades and some memory for our
> customer's SC5832. We lost one blade after 4 years from multiple
> building power fluctuations and have two previously empty slots in the
> chassis. Maybe we'll be able to figure out how to fill the other two
> blade slots, provided we have enough RAM with these new boards.
> The SC5832 has been the MOST reliable SuperComputer we've had on our
> floor. Sure it's slow, but as a user said... "I have over 3000 cores
> just to myself!" They plan on running this machine for as long as
> possible.
> It's really a bummer the company went under. Imagine where they'd be
> today?
yeah it's too bad, but c'est la vie - What's more of a bummer is that it
would be near impossible for another start-up to take a run at solving a
similar problem. If only the government would stop handing out food
stamps to the propreitary vendors and push more grass root/open
initiatives around high efficiency scientific computing. (I don't
necessarily mean low power/frequency, but trying to push the flops/watt
- improving interconnect efficiencies.. cooling. etc)