Tags

seminar instructors get_involved atomic cilk Chip multiprocessor cilk_spawn cilk_for cilk_sync Cilksan commutative operation concurrent agent span critical section data race deadlock cpu determincacy race distributed memory determinism fake lock cilk_scope false sharing global variable hyperobject knot instruction lock linear speedup lock contention multicore nondeterminism multiprocessor mutex nonlocal variable opencilk parallel loop perfect linear speedup race condition process processor receiver reducer parallelism response time scale down running time sequential consistency serial execution serial projection scale up shared memory software performance engineering speedup strand task-parallel sync thread spawn serial semantics throughput worker view events competition work stealing execution time core work