Also known as the Throughput collector. Uses an implementation similar to the Parallel collector, but uses multiple threads to speed up GC. It is best suited to applications with medium- to large-sized data sets.

Parallel compaction allows use use of multiple threads to perform major collections in parallel, improving scalability.

  • Enabled with -XX:+UseParallelGC.
    • Parallel compaction can be disabled with -XX:+UseParallelOldGC.
  • The number of threads is set with -XX:ParallelGCThreads.