Avalanche photodiode gain and impulse response calculation on a public computing platform

Public computing shows promise as an architecture offering tremendous computational power to complex scientific problems. Calculation of the joint probability distribution function for avalanche photodiode gain and impulse response is one such problem. The BOINC architecture is a public computing platform used in this experiment to examine the performance offered by the platform for a specific scientific computation featuring intense computation, task independence, and heavy communication overhead. The large output data of this computation is shown to have a significant impact on the execution of this application resulting in a speedup gain of between 4.5 and 5 times the sequential version. Workload distribution favoring a participant node with rich communication resources in the public computing platform suggests greater performance gains available from the platform if communication resources are increased.