Assessing the Limits of Program-Specific Garbage Collection Performance
We consider the ultimate limits of program-specific garbage collector performance for real programs. We first characterize the GC schedule optimization problem using Markov Decision Processes (MDPs). Based on this characterization, we develop a method of determining, for a given program run and heap size, an optimal schedule of collections for a non-generational collector. We further explore the limits of performance of a generational collector, where it is not feasible to search the space of schedules to prove optimality. Still, we show significant improvements with Least Squares Policy Iteration, a reinforcement learning technique for solving MDPs. We demonstrate that there is considerable promise to reduce garbage collection costs by developing program-specific collection policies.
Fri 17 Jun Times are displayed in time zone: Tijuana, Baja California change
09:00 - 10:00: Memory ManagementResearch Papers at Grand Ballroom Santa Ynez Chair(s): Sam GuyerTufts University | |||
09:00 - 09:30 Talk | Idle Time Garbage Collection Scheduling Research Papers Ulan DegenbaevGoogle, Jochen EisingerGoogle, Manfred ErnstGoogle, Ross McIlroyGoogle, Hannes PayerGoogle Media Attached | ||
09:30 - 10:00 Talk | Assessing the Limits of Program-Specific Garbage Collection Performance Research Papers Nicholas JacekUMass Amherst, Meng-Chieh ChiuUMass Amherst, Benjamin MarlinUMass Amherst, Eliot MossUniversity of Massachusetts Amherst Media Attached |