RFR: JDK-8317987: C2 recompilations cause high memory footprint

Thomas Stuefe stuefe at openjdk.org
Thu Oct 12 14:11:24 UTC 2023


While playing with the new compiler memory statistic (https://github.com/openjdk/jdk/pull/16076), I saw that +StressRecompile causes a twofold increase in memory used during compilations. 

#### Baseline: 

highest peak during compilation is ~60 MB, caused by compilation of `java/lang/invoke/LambdaForm$Kind::<clinit>` with ~18k nodes.

`./images/jdk/bin/java -XX:CompileCommand='collectmemstat,*.*' -XX:-StressRecompilation -Xcomp -Xbatch -cp $REPROS_JAR de.stuefe.repros.Simple`


total     NA        RA        #nodes  time    type  #rc thread              method
(3/1425)
59063K    13406K    40025K    18345   6,174   c2    2   0x00007fdc2c28e7b0  java/lang/invoke/LambdaForm$Kind::<clinit>(()V) 


#### +StressRecompile: 

highest peak during compilation is ~112 MB, same method, sameish node count.

`./images/jdk/bin/java -XX:CompileCommand='collectmemstat,*.*' -XX:+StressRecompilation -Xcomp -Xbatch -cp $REPROS_JAR de.stuefe.repros.Simple`


total     NA        RA        #nodes  time    type  #rc thread              method
(16/1434)
112M      13469K    94987K    18791   16,451  c2    2   0x00007f790428e7b0  java/lang/invoke/LambdaForm$Kind::<clinit>(()V) 


60M -> 112M. In both cases, a large part of the memory was allocated in resource areas. Its relative size compared to the total spike size increased. We accrue a lot of memory in ResourceArea over recompilations.

The solution is to wrap the `Compile` instantiation into a resource mark. Now we are at 61 MB, which is almost baseline:


total     NA        RA        #nodes  time    type  #rc thread              method
(3/1430)
61203K    13469K    40791K    18791   17,013  c2    2   0x00007fdf7428e7b0  java/lang/invoke/LambdaForm$Kind::<clinit>(()V)

-------------

Commit messages:
 - JDK-8317987-C2-recompilations-cause-high-memory-footprint

Changes: https://git.openjdk.org/jdk/pull/16161/files
 Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=16161&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8317987
  Stats: 1 line in 1 file changed: 1 ins; 0 del; 0 mod
  Patch: https://git.openjdk.org/jdk/pull/16161.diff
  Fetch: git fetch https://git.openjdk.org/jdk.git pull/16161/head:pull/16161

PR: https://git.openjdk.org/jdk/pull/16161


More information about the hotspot-compiler-dev mailing list