The number displayed is defined as , for being the maximum number of threads committed with OMP_NUM_THREADS and begin the time the section took with threads.
Efficiency represents scalabilty to many cores. An efficiency close to 100% is desirable, but speedups can happen even with low efficiency. (See speedup)
implies superscalar speedup. In this case, it is likely that one best solution has been found.
Efficiency is affected by multiple parameters. In small programs, removing a parallel section or parallelizing a larger super-section might improve efficiency. Low efficiency can also be caused by contention (). In this case, using StripMining and InLocalStorage/OutLocalStorage might mitigate contention and improve efficiency.