Tommy Ettinger

Normally you don't want to include debugging data in a release JAR, nor do you usually want to go through a JAR build each time you run. That's what `:lwjgl3:run`...

I'm not sure what adding -g and debug would do to a JAR (probably make it a little larger), but for the run task, they don't hurt the performance much,...

RE: removed multiplication, good, that was one of the optimizations I wasn't sure had made it into Fury. I've got a JMH benchmark that can test several different Map implementations...

I was able to run the benchmark, and I may have an avenue that could improve contains() and get() performance.  This was testing just contains() with Object keys, some...

OK, I've been thinking about this for a while, and tinkering a lot with different options. I was surprised to find that at least one version of cuckoo hashing really...

I've done the steps you described, and it has gone from not inlining (code too large) to inlining when hot! I keep finding various small things to change, but they're...

I'm making progress on the rest of the map types now; IdentityMap seems done and ObjectIntMap might be done or close to it. Hybridizing the two shouldn't be hard at...

Well, the `FlipMap` turned out to be significantly slower than all other maps I tried, due to an error I made when writing `put()` and related code. With `put()` fixed...

Speaking only for myself, I was quite pleased to see my name in an `@author Tommy Ettinger` tag in FuryObjectMap. That was part of Kryo initially; some later enhancements to...

Hi @chaokunyang , very nice work on Fury! There's a handful of things I can think of that might speed up maps in your case. An open-addressing, linear-probing table like...