On Fri, 15 Nov 2024 10:05:29 GMT, Aleksey Shipilev <sh...@openjdk.org> wrote:
>> See the bug for more discussion and reproducer. This PR replaces the ad-hoc >> linked list segmented list of arrays. Arrays are easy targets for GC. There >> are possible improvements here, most glaring is parallelism that is >> currently knee-capped by global synchronization. The synchronization scheme >> follows what we have in original code, and I think it is safer to continue >> with it right now. >> >> I'll put performance data in a separate comment. >> >> Additional testing: >> - [x] Original reproducer improves drastically >> - [x] New microbenchmark shows no regression on "churning" tests, which >> covers insertion/removal perf >> - [x] New microbenchmark shows improvement on Full GC times (crude, but >> repeatable), serves as a proxy for reproducer >> - [x] `java/lang/ref` tests in release >> - [x] `all` tests in fastdebug > > Aleksey Shipilev has updated the pull request incrementally with one > additional commit since the last revision: > > Add the node cache Not that it was the main problem here (since using array-linked lists deliver already the major improvement with less changes, kudos!) - but in case having a lock-free structure can be of any help, in JCTools we have and https://github.com/JCTools/JCTools/blob/master/jctools-core/src/main/java/org/jctools/queues/MpmcUnboundedXaddArrayQueue.java which is decently scalable offer side (less consumer-side) and can save reusing chuncks, becoming way simpler. If can be of any interest I can create a gist with a simplified, non-Unsafe version. Otherwise https://github.com/pramalhe/ConcurrencyFreaks/blob/master/Java/com/concurrencyfreaks/queues/array/FAAArrayQueue.java it's as simple as it looks, but probably need to make it linearizable - IDK. ------------- PR Comment: https://git.openjdk.org/jdk/pull/22043#issuecomment-2478475080
