Looming in the cores
summary
Demo shows native threads fail around 17k and apps can spike to 10k–30k threads causing crashes; Project Loom introduces spawn 10,000+ threads OK by adding lightweight virtual threads (fibers) scheduled by JVM implemented with continuations and Java-level schedulers, letting blocking I/O and Thread.sleep become loom-aware. Builds live on jdk.java.net for experimentation but not production; expect libraries and servers (Jetty/Tomcat/Reactor) to adopt loom support so existing synchronous code can run massively concurrent without rewriting.
AI-written summary · about
chapters (20)
- 0:00Introduction and speaker intro
- 1:15About the speaker and Project Loom
- 3:04Why threads matter in modern computing
- 7:18Human vs hardware perspective on threads
- 10:28CPU evolution and multi-core challenges
- 14:23Living in a multi-core world
- 16:33Amdahl's law and parallelism limits
- 21:49Hardware-induced multi-core problems
- 25:38Memory bandwidth and NUMA effects
- 28:05Every system as a distributed system
- 31:54What developers should explore next
- 32:39Thread limitations and Project Loom intro
- 36:36Why blocking code persists
- 39:21Project Loom's lightweight threads
- 43:33How Project Loom works under the hood
- 45:59Where to find Project Loom builds
- 46:42Building multi-core intuition
- 47:45Who will use Project Loom directly
- 48:29Production readiness and migration
- 49:55Project Loom vs reactive frameworks
related talks
51mWhat's Looming in Java The Why and What of Project Loom
35mUntangled: Sneak peek at Project Loom
37mPrepare for what "Loom"s ahead
29mUntangled: Sneak peek at Project Loom
44mLoom: Revolution or Implementation Detail?
01h 01mProject Loom: Revolution in Concurrency or Obscure Implementation Detail?
49mPrepare for what “Loom”s ahead.
01h 13mProject loom: Modern scalable concurrency for the Java platform