Adaptive supercomputing.(LETTERS)(Letter to the editor)

Steve,

With regard to your March 2006 article, great vision! Adaptive Supercomputing does appear to be the path to the future. However, we still have to rely on heroic programming efforts to take advantage of the hardware already on the floor. Even then, we only get a small fraction of peak performance. Combine that with poor scalability, and most applications just cannot take advantage of the homogeneous hardware currently on the floor.

You mention three things that are needed: new compilers; schedulers; and debugging and performance profiling software, run