Increasing the amount of work completed by volunteer computing projects with task distribution policies

Volunteer computing projects rely on volunteers running clients on their computers that contribute to projects when the computers´ owners allow them to. These projects allow people to solve problems that were previously too computationally intensive to solve. However, due to the relatively small fraction of the population that participates in volunteer computing projects, it´s very important to use the donated CPU cycles as efficiently as possible. Volunteer computing clients use two different methods to retrieve tasks: retrieving one task at a time when the client has no more work to do or retrieving multiple tasks at once and storing them in a buffer. We simulate these different task retrieval policies to compare the number of tasks completed by clients using the different policies. Our simulations showed that clients that retrieve one task at a time complete more tasks than clients that retrieve multiple tasks at once and buffer them. Our simulations also showed that there was not a significant gain in the amount of work that could be completed by devising a more complicated adaptive policy.