What JMM Answers
Let’s visualize the Java Memory Model—because in multithreaded code, the big question isn’t just “what did I write?”, it’s “what can another thread actually see, and when?”
Back to gallery
Public AI explainer video
Explain Java memory model with threads visualization
Explain Java memory model with threads visualization
Prompt
Source idea used to generate this video.
Explain Java memory model with threads visualization
Narration
Transcript from the generated scenes.
Threads vs Shared Memory
Picture two threads on the left and right, and in the middle a big box: Shared Memory—think objects on the heap. Each thread also has a little “working copy” area where it can temporarily hold values it read.
Visibility: Stale Reads
Now imagine a shared field x starts at 0 in shared memory. Thread A reads x, gets 0, and keeps that value in its working state. Then A writes x = 1. Here’s the surprise: Thread B might still read x as 0 for a while, because without the right rules, B isn’t guaranteed to see A’s update immediately. That’s the visibility problem.
Ordering: Happens-Before
There’s also ordering. Compilers and CPUs can reorder operations, as long as a single thread would behave the same. So even if A writes two fields, another thread might observe them in a different order, unless we create a happens-before relationship.
More examples