Performance of a Hardware Scheduler for Many-core Architecture

A hardware scheduler for many-core architectures enables fast scheduling and allocation of fine granularity tasks to all cores. We present performance evaluation of a hardware scheduler for HyperCore, a many-core architecture. The evaluation is based on an architectural simulator, using multiple benchmarks representing a wide variety of inherent parallelism. Several architectural improvements are proposed, and various configurations of the scheduler are simulated. The results are analyzed, and are used to highlight the potential and the possible pitfalls of the architecture. It is shown that a scheduler with a capacity to schedule and terminate 10 instances per cycle, along with a task queue of as little as two slots near each core, is sufficient to utilize 256 cores. Other scheduling configurations are also analyzed.