jemalloc vs mallocng benchmarking?

[brainstorm]

I've just noticed that this allocator got switched on by default on MUSL 1.2.1 and will test it soon enough with AWS lambdas (eager to see a 1.0.0 release on musl-cross-make?) :)

Have you run some (rudimentary?) benchmarks vs jemalloc and others to compare?

[brainstorm]

jemalloc bench code: jemalloc/jemalloc@753bbf1

[jvoisin]

Done via daanx/mimalloc-bench#55 :)

[brainstorm]

Cool! have you run those and have some plots/numbers to share? :)

[jvoisin]

Data should show up at https://github.com/daanx/mimalloc-bench/runs/4273418609?check_suite_focus=true in a couple of minutes if everything goes well.

[brainstorm]

According to elapsed_time from https://github.com/daanx/mimalloc-bench/runs/4273418609?check_suite_focus=true#step:5:881 ... malloc-ng performs quite badly compared to the others? :-S

[jvoisin]

Absolutely :/