Okay, this is a shameless plug. It’s my journal. Nyah nyah.
My company has recently hooked up with a couple of corporate partners who own big computers. Said partners want to charge by the CPU-hour for external jobs run on those systems. There are a couple of different models we’re looking at. They range between these options:
- Traditional scientific computing consultant gig: Customer defines a problem, I write code, I run code on the big machine, I write report, customer writes a check, customer gets the report and the code.
- Traditioinal supercomputer center gig: Customer gets a login. Customer gets a bill every month based on usage stats.
We’ve got access to a 2,000 node, quad CPU system, as well as some mid sized (100 – 200 node dual processor). What we need are users.
So, if you’re a scientific / high performance computing person who needs a bigger machine than you have now, and you’d be willing to pay for the CPU cycles … and you’re comfortable being an early adopter (or this sounds like one of your friends) … drop me a line.
Personally, I’m worried about the viability of this business model. That’s why they’re corporate partners and not our very own supercomputers. A lot of companies have failed with the pay for play supercomputing center gig. The difference is that these companies have my group (and other like us in other fields) serving as the user interface.
Leave a Reply