Solving Large Problems Quickly : Progress in 1999

This document describes the progress we have made and the lessons we have learned in 1999 under the NASA Grant NAG2-1230 entitled \Application-Speci c Supercomputing". The long-term goal of this research is to accelerate large, irregular scienti c applications which have enormous data sets and which are di cult to parallelize. To accomplish this goal, we are exploring two complementary techniques: (i) using compilerinserted prefetching to automatically hide the I/O latency of accessing these large data sets from disk; and (ii) using thread-level data speculation to enable the optimistic parallelization of applications despite uncertainty as to whether data dependences exist between the resulting threads which would normally make them unsafe to execute in parallel. Overall, we made signi cant progress in 1999, and the project is going well. This research is supported by NASA Grant NAG2-1230.