Monte Carlo (MC) simulations of photon transport in turbid media suffer a severe limitation represented by very high execution times in all practical cases. This problem could be approached with the technique of parallel computing, which, in principle, is very suitable for MC simulations because they consist in the repeated application of the same calculations to unrelated and superposing events. For the first time in the field of the optical and IR photon transport, we developed a MC parallel code running on the parallel processor computer CRAY-T3E (128 DEC Alpha EV5 nodes, 600 Mflops) at CINECA (Bologna, Italy). The comparison of several single processor runs (on Alpha AXP DEC 2100) and N-processor runs (on Cray T3E) for the same tissue models shows that the computation time is reduced by a factor of about 5*N, where N is the number of used processors. This means a computation time reduction by a factor ranging from about 102 (as in our case) up to about 5*103 (with the most powerful parallel computers) that could make feasible MC simulations till now impracticable.
Application pf parallel computing to a Monte Carlo code for photon transport in turbid media / Colasanti, A.; Guida, G.; Kisslinger, A.; Liuzzi, R.; Quarto, Maria; Riccio, Patrizia; Roberti, G.; Villani, F.. - ELETTRONICO. - 3566:(1998), pp. 73-78. (Intervento presentato al convegno EUROPTO Conference on Photon Propagation in Tissues IV tenutosi a Stockholm nel 9 - 11 settembre 1998) [10.1117/12.334380].
Application pf parallel computing to a Monte Carlo code for photon transport in turbid media
QUARTO, MARIA;RICCIO, PATRIZIA;G. Roberti;
1998
Abstract
Monte Carlo (MC) simulations of photon transport in turbid media suffer a severe limitation represented by very high execution times in all practical cases. This problem could be approached with the technique of parallel computing, which, in principle, is very suitable for MC simulations because they consist in the repeated application of the same calculations to unrelated and superposing events. For the first time in the field of the optical and IR photon transport, we developed a MC parallel code running on the parallel processor computer CRAY-T3E (128 DEC Alpha EV5 nodes, 600 Mflops) at CINECA (Bologna, Italy). The comparison of several single processor runs (on Alpha AXP DEC 2100) and N-processor runs (on Cray T3E) for the same tissue models shows that the computation time is reduced by a factor of about 5*N, where N is the number of used processors. This means a computation time reduction by a factor ranging from about 102 (as in our case) up to about 5*103 (with the most powerful parallel computers) that could make feasible MC simulations till now impracticable.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
