Advanced Topics in Programming Languages: A Lock-Free Hash Table

[@zantrua]

I think if you view the ram as a communication channel with reordering, this state machine thing he's doing is the communicating state machine model from distributed system theory.

[@jwoya]

He also did a talk at Stanford which may be good to watch as well. Check out video WYXgtXWejRM here on KZbin

[@pmcgee003]

live2learn kzbin.info/www/bejne/jYq7mKeOjMqdiK8

[@vaibhavtulsyan7276]

I like that you share the "hash" that uniquely identifies that video here :D

[@nimint3400]

Where sources? :)

[@doug65536]

I completely disagree with the "never shrink" mindset. If you have more than one hash table, having every one permanently the largest size it has ever been quickly becomes a problem. I've seen perf bugs where the customer sees very high memory consumption on a completely idle server, often after the server has been idle for days. If the cost of resize is so cheap you don't allow 'reserve', then why not allow shrink?

[@tohopes]

Why would the server JVM not have enough RAM to handle its maximum expected load?

[@SeanOchoa]

3mo is a long time. Did things work out?