On Wed, 18 Jun 2025 09:00:58 GMT, Albert Mingkun Yang <ay...@openjdk.org> wrote:
>> This patch refines Parallel's sizing strategy to improve overall memory >> management and performance. >> >> The young generation layout has been reconfigured from the previous >> `eden-from/to` arrangement to a new `from/to-eden` order. This new layout >> facilitates young generation resizing, since we perform resizing after a >> successful young GC when all live objects are located at the beginning of >> the young generation. Previously, resizing was often inhibited by live >> objects residing in the middle of the young generation (from-space). The new >> layout is illustrated in `parallelScavengeHeap.hpp`. >> >> `NumberSeq` is now used to track various runtime metrics, such as >> minor/major GC pause durations, promoted/survived bytes after a young GC, >> highest old generation usage, etc. This tracking primarily lives in >> `AdaptiveSizePolicy` and its subclass `PSAdaptiveSizePolicy`. >> >> GC overhead checking, which was previously entangled with adaptive resizing >> logic, has been extracted and is now largely encapsulated in >> `ParallelScavengeHeap::is_gc_overhead_limit_reached`. >> >> ## Performance evaluation >> >> - SPECjvm2008-Compress shows ~8% improvement on Linux/AArch64 and Linux/x64 >> (restoring the regression reported in >> [JDK-8332485](https://bugs.openjdk.org/browse/JDK-8332485) and >> [JDK-8338689](https://bugs.openjdk.org/browse/JDK-8338689)). >> - Fixes the surprising behavior when using a non-default (smaller) value of >> `GCTimeRatio` with Heapothesys/Hyperalloc, as discussed in [this >> thread](https://mail.openjdk.org/pipermail/hotspot-gc-dev/2024-November/050146.html). >> - Performance is mostly neutral across other tested benchmarks: **DaCapo**, >> **SPECjbb2005**, **SPECjbb2015**, **SPECjvm2008**, and **CacheStress**. The >> number of young-gc sometimes goes up a bit and the total heap-size decreases >> a bit, because promotion-size-to-old-gen goes down with the more effective >> eden/survivor-space resizing. >> >> PS: I have opportunistically set the obsolete/expired version to ~~25/26~~ >> 26/27 for now. I will update them accordingly before merging. >> >> Test: tier1-8 > > Albert Mingkun Yang has updated the pull request with a new target base due > to a merge or a rebase. The pull request now contains 19 commits: > > - review > - Merge branch 'master' into pgc-size-policy > - merge > - version > - Merge branch 'master' into pgc-size-policy > - revert-aliases > - Merge branch 'master' into pgc-size-policy > - merge > - merge-fix > - merge > - ... and 9 more: https://git.openjdk.org/jdk/compare/2b94b70e...a21e5363 src/hotspot/share/gc/parallel/parallelArguments.cpp line 71: > 69: // True in product build, since tests using debug build often stress GC > 70: if (FLAG_IS_DEFAULT(UseGCOverheadLimit)) { > 71: FLAG_SET_DEFAULT(UseGCOverheadLimit, trueInProduct); Given only Parallel honors `UseGCOverheadLimit`, you may want to set the default to `trueInProduct` in `gc_globals.hpp` src/hotspot/share/gc/parallel/parallelScavengeHeap.cpp line 381: > 379: HeapWord* result = young_gen()->expand_and_allocate(size); > 380: > 381: if (result == nullptr && !is_tlab && !should_alloc_in_eden(size)) { I feel that you changed intent of `should_alloc_in_eden()`. I believe the original intent is to prevent allocating large objects in eden, and here, seems to prevent allocating small objects in old gen. What benefits do you get? ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/25000#discussion_r2160152953 PR Review Comment: https://git.openjdk.org/jdk/pull/25000#discussion_r2160163706
