G1 GC NUMA-Aware

https://ionutbalosin.com/2020/01/hotspot-jvm-performance-tuning-guidelines/

  • -XX:+UseNUMA (default false)

Improve the performance of a JVM that runs on a machine with nonuniform memory architecture (NUMA), or multiple sockets, by enabling the NUMA aware Collector to allocate objects in memory node local to a processor, increasing the application’s use of lower latency memory.

Note: at the moment the NUMA aware Collectors are: Parallel GC, G1 GC, and ZGC.

G1 GC NUMA-Aware#

https://sangheon.github.io/2020/11/03/g1-numa.html

The default Garbage Collector, G1 GC was enhanced on JDK-14 by making its memory allocator NUMA-aware by JEP-345: NUMA-Aware Memory Allocation for G1 ^1^. In this article, I will explain a little bit about its implementation. I will give minimal explanations about how G1 GC works in general.

For Java 8 G1 GC#

https://www.mail-archive.com/hotspot-gc-use@openjdk.java.net/msg00390.html

G1 is not NUMA aware. Only Parallel GC(-XX:+UseParallelGC) has NUMA awareness.

May I use the -XX:+UseNUMA flag with -XX:+UseG1GC?

Yes, you can use +UseNUMA with +UseG1GC. G1 algorithm is not NUMA aware, but with UseNUMA enabled, we can get NUMA interleaving. i.e. Interleave memory across NUMA nodes if possible and this would give some benefits on NUMA system. FYI, UseNUMAInterleaving(false by default) command-line option will be automatically enabled if UseNUMA is enabled.

https://docs.oracle.com/javase/7/docs/technotes/guides/vm/performance-enhancements-7.html The NUMA-aware allocator can be turned on with the -XX:+UseNUMA flag in conjunction with the selection of the Parallel Scavenger garbage collector. The Parallel Scavenger garbage collector is the default for a server-class machine. The Parallel Scavenger garbage collector can also be turned on explicitly by specifying the -XX:+UseParallelGC option.

parallel collector NUMA-Aware#

https://bugs.openjdk.java.net/browse/JDK-8210473

The parallel collector, enabled by by -XX:+UseParallelGC, has been NUMA-aware for many years. This has helped to improve the performance of configurations that run a single JVM across multiple sockets.

Ref#

https://frankdenneman.nl/2015/02/20/memory-deep-dive/