Gerry Creager wrote:
SiCortex is not too far from the Beowulf concept but is possibly closer to a packaged single DM entity. I'd thought of it as a low-power, low heat (DM) HPC system without ever considering the beowulfness of it. It uses Linux and MPI, so it meets my needs in that regard. It's low-power and thus lower heat; these are good. It's slower than current PC class hardware. Bad selection for me at this point.
I think there is a SiCortex person on the list somewhere. My question is how is this fundamentally different in the approach than Orion Multisystems? Yes, I know about the technological differences. The issue is the approach to market: lower power (in all senses of the word) CPUs coupled together.
For them to be successful, they need customers to buy machines, which usually means getting commercial codes onto them. This is *hard*. It cannot be understated how hard that is for a new architecture. You have to start by convincing the ISV that their code will generate more revenue for the ISV than cost. Then you have to convince the customer to buy.
Orion never really hit the critical mass it needed to convince major ISVs to buy into it and target this as a platform to deliver cycles. Most of the ISVs I have spoken to recently are looking at acceleration technologies for their needs. Remember, the ISVs want end users to spend less on the hardware so that the users will feel like they are spending less in general, even if the software price is the same or more. How Orion helped with that is unknown. How SiCortex helps with this is also unknown.
Will be interesting to see. Hopefully more than just a few researchers will buy their gear.
-- Joseph Landman, Ph.D Founder and CEO Scalable Informatics LLC, email: [EMAIL PROTECTED] web : http://www.scalableinformatics.com http://jackrabbit.scalableinformatics.com phone: +1 734 786 8423 fax : +1 866 888 3112 cell : +1 734 612 4615 _______________________________________________ Beowulf mailing list, Beowulf@beowulf.org To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
