Abstract:

A configurable simulation environment has been developed to evaluate task‐scheduling strategies in multicore systems with complex inter‐task dependencies. The framework comprises a Configuration Loader, Main Controller, Test Runner, Task Generator, DepThread Class, and DepRunner Manager, and supports both YAML‐based and random parameterization of execution‐time distributions, task‐arrival dynamics, CPU‐affinity probabilities, and dependency densities with enforced acyclic graphs. Each task executes a compute‐bound workload to impose real CPU load. Two execution models are compared: a standard scheduler that launches tasks immediately and allows passive waiting, and a dependency‐aware model (DAM) that defers activation until all prerequisites are met. Metrics are recorded in CSV format and visualized via a matplotlib‐generated bar chart showing percentage improvement alongside dependency counts. A representative experiment demonstrates that DAM reduces idle processor occupation and yields significant execution‐time gains in dependency‐rich scenarios.