Search

General purpose garbage collectors have yet to combine short pause times with high throughput. For example, generational collectors can achieve high throughput. They have modest average pause times, but occasionally collect the whole heap and consequently incur long pauses. At the other extreme, concurrent collectors, including reference counting, attain short pause times but with significant performance penalties. This paper introduces a new hybrid collector that combines copying generational collection for the young objects and reference counting the old objects to achieve both goals. It restricts copying and reference counting to the object demographics for which they perform well. Key to our algorithm is a generalization of deferred reference counting we call Ulterior Reference Counting. Ulterior reference counting safely ignores mutations to select heap objects. We compare a generational reference counting hybrid with pure reference counting, pure mark-sweep, and hybrid generational mark-sweep collectors. This new collector combines excellent throughput, matching a high performance generational mark-sweep hybrid, with low maximum pause times.

In this paper, we describe the design, implementation, and evaluation of a new garbage collection framework called the Java Memory Management Toolkit (JMTk). The goals of JMTk are to provide an efficient, composable, extensible, and portable toolkit for quickly building and evaluating new and existing garbage collection algorithms. Our design clearly demarcates the external interface between the collector and the compiler for portability. For extensibility, JMTk provides a selection of allocators, garbage identification, collection, pointer tracking, and other mechanisms that are efficient and that a wide variety garbage collection algorithms can compose and share. For instance, our mark-sweep and reference counting collectors share the free list implementation. We perform a comprehensive and detailed study of collectors including copying, mark-sweep, reference counting, copying generational, and hybrid generational collectors using JMTk in Jikes RVM on a uniprocessor. We find that the performance of collectors in JMTk is comparable to the highly tuned original Jikes collectors. In a study of full heap and generational collectors, we confirm the significant benefits of generational collectors on a wide variety of heap sizes, and reveal that on very small heaps, collection time is enormous. These experiments add other new insights, such as firmly establishing that for a variety of generational collectors, a variable-size nursery which is allowed to grow to fill all the space not in use by the older generation performs better than a fixed-size nursery. We thus show the utility of extensive collector comparisons and establish the benefits of our design.

General purpose garbage collectors have yet to combine short pause times with fast throughput. Collectors such as reference counting attain short pause times with significant performance penalties. Pure and hybrid generational collectors can achieve high throughput and have modest average pause times, but occasionally collect the whole heap and consequently incur long pauses. This paper introduces RC-hybrid, which combines copying nursery collection and reference counting the older generation to achieve both goals. Key to our algorithm is a generalization of deferred reference counting which allows RC-hybrid to safely ignore mutations to nursery objects. RC-hybrid thus restricts copying and reference counting to the objects for which they perform well. Copying collectors’ bump pointer allocation is extremely fast, and collection time is proportional to the live objects whose survival rates are low and bounded in a fixed size nursery. Reference counting time is proportional to object mutations and the number of dead objects, both of which are typically low for older objects. We further bound time spent reference counting with collection triggers and buffering. We compare RC-hybrid with pure reference counting, a stateof-the-art copying and mark-sweep hybrid, and a number of other collectors. We show that RC-hybrid combines fast throughput, and low average and maximum pause times.

Until recently, the best performing copying garbage collectors used a generational policy which repeatedly collects the very youngest objects, copies any survivors to an older space, and then infrequently collects the older space. A previous study that used garbage-collection simulation pointed to potential improvements by using an Older-First copying garbage collection algorithm. The Older-First algorithm sweeps a fixed-sized window through the heap from older to younger objects, and avoids copying the very youngest objects which have not yet had sufficient time to die. We describe and examine here an implementation of the Older-First algorithm in the Jikes RVM for Java. This investigation shows that Older-First can perform as well as the simulation results suggested, and greatly improves total program performance when compared to using a fixed- the nursery generation collector. We further compare Older-First to a flexible-size nursery generational collector in which the nursery occupies all of the heap that does not contain older objects. In these comparisons, the flexible-nursery collector is occasionally the better of the two, but on average the Older-First collector performs the best.

We present the design and implementation of a new garbage collection framework that significantly generalizes existing copying collectors. The Beltway framework exploits and separates object age and incrementality. It groups objects in one or more increments on queues called belts, collects belts independently, and collects increments on a belt in first-in-first-out order. We show that Beltway configurations, selected by command line options, act and perform the same as semi-space, generational, and older-first collectors, and encompass all previous copying collectors of which we are aware. The increasing reliance on garbage collected languages such as Java requires that the collector perform well. We show that the generality of Beltway enables us to design and implement new collectors that are robust to variations in heap size and improve total execution time over the best generational copying collectors of which we are aware by up to 40%, and on average by 5 to 10%, for small to moderate heap sizes. New garbage collection algorithms are rare, and yet we define not just one, but a new family of collectors that subsumes previous work. This generality enables us to explore a larger design space and build better collectors.

In many garbage collected systems, the mutator performs a write barrier for every pointer update. Using generational garbage collectors, we study in depth three code placement options for remembered-set write barriers: inlined, out-of-line, and partially

We present the design and implementation of a new garbage collection framework that significantly generalizes existing copying collectors. The Beltway framework exploits and separates object age and incrementality. It groups objects in one or more increments on queues called belts, collects belts independently, and collects increments on a belt in first-in-first-out order. We show that Beltway configurations, selected by command line options, act and perform the same as semi-space, generational, and older-first collectors, and encompass all previous copying collectors of which we are aware. The increasing reliance on garbage collected languages such as Java requires that the collector perform well. We show that the generality of Beltway enables us to design and implement new collectors that are robust to variations in heap size and improve total execution time over the best generational copying collectors of which we are aware by up to 40%, and on average by 5 to 10%, for small to moderate heap sizes. New garbage collection algorithms are rare, and yet we define not just one, but a new family of collectors that subsumes previous work. This generality enables us to explore a larger design space and build better collectors.

Pretenuring can reduce copying costs in garbage collectors by allocating long-lived objects into regions that the garbage collector will rarely, if ever, collect. We extend previous work on pretenuring as follows. (1) We produce pretenuring advice that is neutral with respect to the garbage collector algorithm and configuration. We thus can and do combine advice from different applications. We find that predictions using object lifetimes at each allocation site in Java programs are accurate, which simplifies the pretenuring implementation. (2) We gather and apply advice to applications and the Jalapeño JVM, a compiler and run-time system for Java written in Java. Our results demonstrate that building combined advice into Jalapeño from different application executions improves performance regardless of the application Jalapeño is compiling and executing. This build-time advice thus gives user applications some benefits of pretenuring without any application profiling. No previous work pretenures in the run-time system. (3) We find that application-only advice also improves performance, but that the combination of build-time and application-specific advice is almost always noticeably better. (4) Our same advice improves the performance of generational and Older First collection, illustrating that it is collector neutral.