RFR: Mark-compact performance improvements and cleanups

[Zhengyu Gu]

Good to me.

-Zhengyu

On 05/29/2018 03:20 PM, Aleksey Shipilev wrote:
> http://cr.openjdk.java.net/~shade/shenandoah/mark-compact-improvs/webrev.01/
> 
> This is the pack of mark-compact cleanups and improvements.
> 
> *) mc-logging: More verbose profiling for phase 4 in mark-compact
> 
> Adds a few more profiling points in "Copy object" phase of mark compact, which enables better
> performance diagnostics.
> 
> *) mc-liveness: Full GC always comes with liveness data
> 
> We used to skip liveness counting for mark-compact, because the core algorithm does not require it.
> It provided some boosts when our liveness calculation was a major performance hog. It is not
> anymore, and having liveness data enables us to do shortcuts, like below. This also cleans up shared
> code substantially.
> 
> *) mc-immediate-trash: Reclaim immediate garbage after mark-compact marking
> 
> Once we know liveness, we can reclaim immediate garbage regions right after marking. This lets
> threads down below treat the regions as the targets for compaction immediately, instead of
> compacting through them. This is especially important for figuring out compaction for humongous
> regions, which is planned before the regular object moves.
> 
> *) mc-shortcut-live: Shortcut regions that are known not to be alive
> 
> Same here, liveness provides us with insight if we even want to walk marked objects in the region,
> or not. For the fully-garbage region it would be futile, and it will yield no marked objects.
> 
> Motivational performance data:
> 
>   - Serial: 8191M->4M Full GC cycles
> 
>      before:
>        Pause Full GC (N) = 0.90 s (a =    64425 us) (n =    14)
>               (lvls, us =    56250,    56836,    57617,    58594,   103983)
> 
>      after:
>        Pause Full GC (N) = 0.22 s (a =    14914 us) (n =    15)
>               (lvls, us =    12891,    13281,    14453,    15820,    16818)
> 
>   - retain.tree: 10239M->1070M Full GC cycles
> 
>      before:
>        Pause Full GC (N) = 3.50 s (a =   291548 us) (n =    12)
>               (lvls, us =   275391,   281250,   287109,   296875,   314651)
> 
>      after:
>        Pause Full GC (N) = 3.54 s (a =   294925 us) (n =    12)
>               (lvls, us =   285156,   287109,   292969,   296875,   304701)
> 
>   - LRUFragger: 102138M->65687M Full GC cycles
> 
>      before:
>        Pause Full GC (N) = 151.24 s (a =  4726195 us) (n =    32)
>               (lvls, us =  4570312,  4648438,  4707031,  4765625,  4889809)
> 
>      after:
>        Pause Full GC (N) = 153.10 s (a =  4784420 us) (n =    32)
>               (lvls, us =  4667969,  4746094,  4765625,  4804688,  4929369)
> 
> Testing: hotspot_gc_shenandoah {fastdebug|release}, benchmarks
> 
> Thanks,
> -Aleksey
>