Earliest published evidence that instruction level parallelism was practically exploitable in ordinary programs. Developed a set of parallelizing transformations that could deal with unpredictable control flow in programs and produce provably superior parallel code (Percolation Scheduling). With A. Aiken developed a software pipelining technique combining general handling of control-flow with optimal results (Optimal Loop Parallelization). These techniques have been used at IBM (Watson), Siemens (Munich) and Fujitsu Labs (Japan), and in GCC. With L. Hendren and Joe Hummel developed pointer analysis techniques for complex data-structures. Contributed to Design Automation by designing transformations for Systhesis. Developed new hybrid algorithms for fast Matrix Multiplication, benchmarking metodology for WebRTC, new light-weight synchronization and scheduling techniques for parallel machines. Authored/Co-authored over 300 peer-reviewed papers and several books.