I've been doing some performance testing as part of the work on the new core design and noticed thread contention when using chunks due to ClassTag lookups, which are synchronized by Scala's ClassTag cache. We don't really need the ClassTag lookup since the goal in chunks is to always use Any for internal arrays. I've tried to introduce a benchmark for it but I couldn't reproduce the issue in isolation.
I'm updating both versions in kyo-core and kyo-prelude.
I've been doing some performance testing as part of the work on the new core design and noticed thread contention when using chunks due to
ClassTag
lookups, which are synchronized by Scala'sClassTag
cache. We don't really need theClassTag
lookup since the goal in chunks is to always useAny
for internal arrays. I've tried to introduce a benchmark for it but I couldn't reproduce the issue in isolation.I'm updating both versions in
kyo-core
andkyo-prelude
.