The cost of building the thing has already been paid. It is a sunk cost. If it costs $352/hour to run the roadrunner and they can only get $150/hour in income to operate it, it don't matter how much it cost to build it.
If it costs $352/hour to run the roadrunner and they can only get $150/hour in income to operate it, it don't matter how much it cost to build it.
Assuming they already paid for the replacement super-computer too etc. Otherwise, sure it matters. Instead of finding $xx-$xxx million to build a more efficient one, you might be better off to pay the electricity bill difference. This is still in the top 25 computers in the world, so it can be used for a lot of things, but we don't know all the details.
It's a distributed batch-job type system, not a single computer with a petaflop core. I don't think it is technically accurate to call it a cluster, but it isn't one machine either.
Unless you've got a big HPC job to run, it wouldn't be very exciting. You log in to a "login node" which is just an ordinary linux server and you get a bash prompt. Nothing very remarkable. To actually do anything you would create a script and submit your job to the workload manager, then come back later for your results.
