Research on Algorithm of Parallel Garbage Collection Based on LISP 2 for Multi-core System

As the multi-core architecture prevails, garbage collection is becoming an essential feature to support high performance systems. In this paper, we have described a parallel copying garbage collector and presented experimental results that demonstrate good scalability. The collector is designed for shared-memory multi-core systems or multiprocessor systems in the base of classical LISP2 algorithm. The earlier parallel copying collector can not compact the live data into a single contiguous region at one end of the heap, but leaves multiple object groups, one for every two neighboring partitions. Our novel algorithm overcomes this drawback. The average collector speedup is 1.65 at two-core system and 2.4 at four-core system.