JEP 522 performance regression with large pages

[Thomas Schatzl]

Hi Brian,

On 28.08.25 16:42, Brian S O'Neill wrote:
 > I'm experimenting with the changes in PR 23739 (8342382: Implement JEP
 > 522: G1 GC: Improve Throughput by Reducing Synchronization) and I'm
 > seeing a small performance regression when large pages are configured.
 >

   thanks for giving the change a spin.

 > The test is fairly complicated, and most of the memory it uses is off
 > heap. The GC heap size is set to 3GB (min and max), which is much
 > larger than is actually required. A bunch of objects are allocated up
 > front and remain in the old gen for the duration of the test run.
 > Between each GC cycle, almost all the old gen objects will have been
 > updated to reference a young object. The young gen objects live for
 > about 2 microseconds, and the references from the old gen objects are
 > cleared.
 >
 > Here's the baseline results when running with "normal" pages:
 >
 > ParallelGC:   235.6 seconds
 > G1GC JEP 522: 238.7 seconds
 > G1GC:         241.5 seconds
 > ZGC:          246.2 seconds
 >
 > With JEP 522, there's a small performance improvement, about 1%, which
 > is nice to see. Here's the results when running with large pages (-XX:
 > +UseLargePages -XX:+UseTransparentHugePages shmem_enabled is advise):

Just to make sure it's not THP: can you also try with pre-allocated 
large pages?

E.g. via
sysctl -w vm.nr_hugepages=<nr>

and

-XX:-UseTransparentHugePages?

 >
 > ParallelGC:   228.9 seconds
 > G1GC:         235.1 seconds
 > ZGC:          239.3 seconds
 > G1GC JEP 522: 239.7 seconds
 >
 > All of the GCs show a performance improvement when using large pages,
 > but with JEP 522, G1 is slower than the current version (JDK 24).
 >
 > I don't know why there's a performance regression. Is this to be
 > expected with large pages, or is there a missing configuration
 > somewhere? I'm not configuring anything other than -Xms, -Xmx, and the
 > large page settings. Also note that the test is run ten times (without

Can you give the full command line?

 > restarting the JVM) and the average time is reported.
 >

No, we are not aware of such a regression. We do run benchmarks with 
large pages regularly, although not so often with THP but pre-allocated 
ones.

Thanks,
   Thomas