the petaflop barrier broken!

IBM has just made public Blue Gene/P, a 294912 processor machine that claims a 1 petaflop performance that I am sure will be used to its full extent by the US Department of Energy (DOE) at Argonne National Lab (Illinois). Now that the US is going to have a petaflop facility I wonder how long it will take Japan to overtake them...

Speaking of which, the code I developed to run in Earth Simulator (well, the non-vectorial version of it) will soon be running on the Blue Gene/W at IBM's T.J. Watson Center, and I am looking forward to see the scaling over thousands of processors.

Hmm... at some point I should stop working on the code efficiency and focus on the physics of the problem... What can I say? I do like to play with code to improve its performance :)

Mr. C