The next-generation supercomputer project and a plan for the advanced institute for computational science

Computer simulation has become the critical third pillar for scientific discovery along with experiment and theory. We are running the next-generation supercomputer project to design, build and set up the supercomputer, the world's most advanced computer with a speed of LINPACK target of 10 peta flops. The next-generation supercomputer will be available in three years (by 2012). With the emergence of petascale computing platforms we are entering a new era of modeling. Establishment of the COE in the field of supercomputing is also being planned. We would like to create a new, independent, international COE of computer science and computational science in 2010, open to individuals from all over the world from different backgrounds. Petascale computing will enable us to simulate physical processes on a scale never seen before, and approach convergence for dynamical processes never thought possible. Petascale computing will open up a new frontiers. Its success will have lasting impact on science and technology and people all around the world and for generations into the future. An overview of the project and the most recent progress will be discussed.