High-performance computing and large applications in computational algebra
Professor Cooperman researches high-performance computing and large applications in computational algebra. He developed Task-Oriented Parallel C (TOP-C/C++), a model for writing parallel software easily. More recently, his research concerns novel applications of transparent checkpointing, such as checkpointing symbolic debuggers and checkpointing the graphics desktop at the process level. His DMTCP checkpointing project provides a robust platform for this purpose, which does not require any kernel modules or kernel patches. His joint work on disk-based parallel computing, including the novel language Roomy, provides a traditional sequential programming API, which transparently makes use of many terabytes of parallel disk storage. Finally, the problem of converting many sequential processes into a single multi-threaded process is an important focus of his ongoing collaboration with CERN (European Center for Nuclear Research), and that work is motivated by the needs for future many-core CPU chips.