"High Performance Clojure"
summary
Research shows a new chard CSV/JSON parser (zero dependencies) that parses via a raw character buffer and canonicalizes keys to reuse String hash codes, beating data.json and jsonista in benchmarks. A library named ham-fisted persistent hash map implements bitmap-trie and flat power-of-two hash tables, supports persistent↔transient transitions and copy-on-write arrays, and moves primitives like computeIfAbsent and in-map union into the DS. That yields 5–10× faster frequencies and set unions; sort uses fastutil and some stream double-array cases show >20× gains versus current Clojure approaches.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 6:38Hamfisted
- 12:30Graphs
- 20:40Monte Carlo
- 28:15Binning
- 29:02frequencies
- 30:30wrapping up
- 33:15conclusion
related talks
01h 12mHidden pearls for high-performance-persistence in Java
32mClojure, Java 9, and You
48mFunctional data structures in Java
29mPacket Capturing with Clojure
51mJava at Speed: Getting the Most Out of Modern Hardware
41mJava 24, Faster Than Ever
49m1BRC – Nerd Sniping the Java Community
52mMicro-Optimization for Fun and Profit