Bacon01Java.pptVIP

Java without the Coffee Breaks:A Nonintrusive MultiprocessorGarbage Collector David F. Bacon IBM T.J. Watson Research Center Joint work with C.R. Attanasio, Han Lee, V.T. Rajan, and Steve Smith Combines results to appear in PLDI01 and ECOOP’01 The Garbage Collector: What is It? Concurrent Multiprocessor Reference-counted Cycle collecting Low-latency High-performance Why do it? RC has good locality properties Will tracing collection scale to multi-GB heaps? Effect of rising memory latency, SMT/CMP? RC is easily decoupled from mutator Promises lower synchronization costs Java makes good GC a commercial requirement Makes sense to have “genetic diversity” People said it couldn’t be done Why is it hard? Must defer reference counting stack Complicates algorithm Reference counts are shared variables Synchronization required Write barriers for RC more expensive Affect two objects RC doesn’t handle cycles Other systems use backup tracing collector Outline Introduction Motivation System Overview Concurrent Reference Counting Cycle Collection Measurements Conclusions System Overview Producer/Consumer System Similar to Deutsch-Bobrow DRC As implemented by DeTreville on SRC Firefly Implementation Implemented in Jalape?o JVM at IBM TJW All of VM, JIT, and GC are written in Java Extended with unsafe primitives Multiple GC implementations Runs on IBM RS/6000 multiprocessors Linux/Intel port underway GC is machine-independent except for barriers Outline Introduction Motivation System Overview Concurrent Reference Counting Cycle Collection Measurements Conclusions Concurrent Reference Counting Time divided into epochs All CPUs must participate before epoch advances Write barrier on heap updates inc/dec operations placed in buffer Objects allocated with RC=1, dec enqueued Decrements processed one epoch behind increments Stack references are deferred Snapshot stacks at epoch boundary First increment; decrement at next epoch Simpler invariant than Deutsch-Bobrow; no ZCT r