IBM User Guide for Java V7 on Windows > IBM SDK for Java > The collector and the allocator > Balanced Garbage Collection policy

NUMA awareness

The Balanced Garbage Collection policy can increase application performance on large systems that have Non-Uniform Memory Architecture (NUMA) characteristics.

NUMA is used in multiprocessor systems on x86 and IBM POWER architecture platforms. In a system that has NUMA characteristics, each processor has local memory available, but can access memory assigned to other processors. The memory access time is faster for local memory. A NUMA node is a collection of processors and memory that are mutually close. Memory access times within a node are faster than outside of a node.

The Balanced Garbage Collection policy can split the Java™ heap across NUMA nodes in a system. Application threads are segregated such that each thread has a node where the thread runs and favors the allocation of objects. This process increases the frequency of local memory access, increasing overall application performance.

A Partial Garbage Collection (PGC) attempts to move objects closer to the objects and threads that refer to them. In this way, the working set for a thread is physically close to where it is running.

The segregation of threads is expected to improve application performance. However, there might be some situations where thread segregation can limit the ability of an application to saturate all processors. This issue can result in slight fragmentation, slowing performance. You can test whether this optimization is negatively affecting your application by turning off NUMA awareness using the -Xnuma:none command-line option.

Parent: Balanced Garbage Collection policy

Error 404 - Not Found

The document you are looking for may have been removed or re-named. Please contact the web site owner for further assistance.