Work-Stealing Scheduler: Parallel Task Load Balancing

Parallel computing splits one big job into many independent tasks — but handing those tasks out efficiently is its own algorithmic problem. This simulation renders a bank of worker cores, each holding its own local task deque as a stack of glowing blocks. Every core pops and executes its own tasks at its own pace; the moment a core's stack runs dry it becomes a thief, picks another core at random, and steals a task straight off the bottom of that core's deque — visualised as an arcing steal event between columns. Tune the core count, the total task pool and the average task duration, flip stealing on or off to see cores starve instead, and watch throughput, steal count and load imbalance update live as the run drains to completion.