nfsantos commented on PR #2178: URL: https://github.com/apache/jackrabbit-oak/pull/2178#issuecomment-2723842652
> @thomasmueller @nfsantos Please find the benchmark below: > > Each method creates a Set from Iterables and iterates over it. So far I haven't gotten any OOM from either of them. > > > Benchmark (size) Mode Cnt Score Error Units > > ImmutableSet.apache 1 avgt 10 0.013 ± 0.001 us/op > > ImmutableSet.apache 100 avgt 10 1.171 ± 0.006 us/op > > ImmutableSet.apache 10000 avgt 10 112.914 ± 7.751 us/op > > ImmutableSet.apache 1000000 avgt 10 22179.836 ± 3071.103 us/op > > ImmutableSet.apache 100000000 avgt 10 3063667.217 ± 70743.291 us/op > > ImmutableSet.guava 1 avgt 10 0.001 ± 0.001 us/op > > ImmutableSet.guava 100 avgt 10 1.026 ± 0.006 us/op > > ImmutableSet.guava 10000 avgt 10 128.871 ± 9.784 us/op > > ImmutableSet.guava 1000000 avgt 10 51641.190 ± 2674.666 us/op > > ImmutableSet.guava 100000000 avgt 10 3225314.017 ± 661972.699 us/op > > ImmutableSet.jdk 1 avgt 10 0.017 ± 0.001 us/op > > ImmutableSet.jdk 100 avgt 10 1.306 ± 0.022 us/op > > ImmutableSet.jdk 10000 avgt 10 113.677 ± 6.856 us/op > > ImmutableSet.jdk 1000000 avgt 10 24777.681 ± 9976.868 us/op > > ImmutableSet.jdk 100000000 avgt 10 3145239.833 ± 121858.887 us/op Thanks for the benchmarks. Could you do similar benchmarks for lookups? Fast lookups are one of the main reasons to use a set instead of a list, so I would imagine that most Sets created by Oak will be heavily used for lookups. Therefore I think lookup performance is more important than traversal performance. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
