RADICAL-Pilot and Parsl: Executing Heterogeneous Workflows on HPC Platforms

TL;DR

This paper discusses integrating RADICAL-Pilot and Parsl to enable efficient execution of heterogeneous workflows on HPC platforms, enhancing development and scalability for scientific applications.

Contribution

It presents a novel integration approach of RADICAL-Pilot and Parsl, including the development of RPEX, to support heterogeneous tasks on HPC resources.

Findings

RPEX enables execution of heterogeneous MPI Python functions on CPU and GPU resources.

The integrated system demonstrates scalable performance for scientific workflows.

The approach improves workflow development efficiency and resource utilization.

Abstract

Workflows applications are becoming increasingly important to support scientific discovery. That is leading to a proliferation of workflow management systems and, thus, to a fragmented software ecosystem. Integration among existing workflow tools can improve development efficiency and, ultimately, increase the sustainability of scientific workflow software. We describe our experience with integrating RADICAL-Pilot (RP) and Parsl as a way to enable users to develop and execute workflow applications with heterogeneous tasks on heterogeneous high-performance computing resources. We describe our approach to the integration of the two systems and detail the development of RPEX, a Parsl executor which uses RP as its workload manager. We develop an RP executor that executes heterogeneous MPI Python functions on CPU cores and GPUs. We measure the weak and strong scaling of RPEX, RP, and Parsl when providing new capabilities to two paradigmatic use cases: Colmena and Ice Wedge Polygons