Scaling PPE beyond dual-core and to heterogeneous systems

Extend and experimentally evaluate the Partition and Parallel Execution (PPE) runtime architecture on higher core counts and heterogeneous hardware systems to ascertain effective scheduling strategies and partition granularity across diverse processor and accelerator configurations.

Background

PPE is introduced as a runtime architecture enabling parallel execution by partitioning workloads across cores, demonstrated primarily with dual-core experiments. To validate generality and scalability, the authors highlight the need to extend PPE to many-core and heterogeneous environments.

This open direction targets understanding scheduling and partitioning trade-offs across varying hardware, which is crucial for performance predictability and utilization on modern multi-core and heterogeneous platforms.