Tags

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