search

eishay / jvm-serializers

Benchmark comparing serialization libraries on the JVM
http://groups.google.com/group/java-serialization-benchmarking
3.29k stars 562 forks source link

delete (dealloc) time missing from performance wiki #74

Closed KjellKod closed 3 years ago

KjellKod commented 7 years ago

Hi,

the create time is visible in the performance breakdown but not the deallocation time. For applications which huge number of object creation the object deletion might be just as important for the performance.

cakoose commented 7 years ago

I don't know a good way to measure deallocation time with garbage collected memory.

For manually managed (or ref-counted) memory, deallocation time is proportional to the amount of memory being deallocated. For garbage collected memory, the deallocation time is affected by the amount of live memory as well, which makes it harder to measure.

One idea is to run the JVM with a small heap, to make sure garbage collection would be triggered multiple times in a single test run. This way, deallocation time is included in the overall running time measurement.

Do identify the deallocation time, maybe you could also run the benchmark with a large heap and see what the difference is. I'm not sure how precise this would be.

On Fri, Apr 7, 2017 at 2:42 PM, Kjell Hedström notifications@github.com wrote:

Hi,

the create time is visible in the performance breakdown but not the deallocation time. For applications which huge number of object creation the object deletion might be just as important for the performance.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/eishay/jvm-serializers/issues/74, or mute the thread https://github.com/notifications/unsubscribe-auth/AAOpdb9bDqMBwqjjBmmaXLfeyxOZjj_Dks5rtq3YgaJpZM4M3bDK .