Hi, You can log reference processing counts and times… If you’re at about 50K references/second then using parallel reference processing starts to make sense… prior, single threaded is better.
Kind regards, Kirk > On Jan 29, 2019, at 8:56 AM, Nat <[email protected]> wrote: > > Recently, I found a situation where -XX:+ParallelRefProcEnabled makes the GC > time to be significantly worse, especially the tail end. It seems to be much > worse with multiple core system. I believe it is a bug that might not be > addressed yet. The key problem seems to be around the task queue > implementation rather than the actual work itself. So instead of just > enabling it blindly, make sure you cut and measure. > > I will update if I find more information. > > -- > You received this message because you are subscribed to the Google Groups > "mechanical-sympathy" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected] > <mailto:[email protected]>. > For more options, visit https://groups.google.com/d/optout > <https://groups.google.com/d/optout>. -- You received this message because you are subscribed to the Google Groups "mechanical-sympathy" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
