Title :
Assessing parallel algorithms
Author :
Baskakov, Yuriy ; Golubev, Ivan
Author_Institution :
St. Petersburg Electrotech. Univ. "LETI", St. Petersburg, Russia
Abstract :
Various existing performance metrics within the parallel systems domain are analyzed. These include the different flavors of speedup, efficiency, and isoefficiency. Execution time still remains the most widely used metric. A new method to automatically estimate algorithmic cost is provided.
Keywords :
computational complexity; parallel algorithms; algorithmic complexity; algorithmic cost estimation; isoefficiency; parallel algorithms; parallel systems domain; performance metrics; Algorithm design and analysis; Complexity theory; Estimation; Hardware; Measurement; Parallel algorithms; Software algorithms; algorithmic complexity; algorithmic cost models; automatic algorithmic cost estimation; parallel computing;
Conference_Titel :
Soft Computing and Measurements (SCM), 2015 XVIII International Conference on
Conference_Location :
St. Petersburg
Print_ISBN :
978-1-4673-6960-2
DOI :
10.1109/SCM.2015.7190460
