Task-Based Programming for Adaptive Mesh Refinement in Compressible Flow Simulations

2025-08-07

AI Summary

High-order compressible flow solvers suffer from inefficient AMR dynamic mesh management, high task-scheduling overhead, and insufficient GPU programming abstractions. Method: This paper proposes three key techniques built upon the Regent language and Legion runtime: (1) a task-fusion strategy tailored to AMR’s dynamic block-structured hierarchy to reduce fine-grained task launch overhead; (2) a lightweight annotation-driven mechanism for automatic GPU kernel generation, improving performance and simplifying heterogeneous programming; and (3) an AMR data structure jointly optimized for mesh topological validity and load balance. Contribution/Results: Experiments demonstrate an 18× speedup in task scheduling due to fusion; automatically generated GPU kernels achieve 9.7× higher performance than hand-written counterparts; and the approach successfully simulates two canonical Euler equation test cases—vortex advection and shock–vortex interaction—validating its efficiency and practicality for production-scale high-order CFD.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Hardware-aware ML
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Data management and stream processing for Web, mobile and wireless applications
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

High-order solvers for compressible flows are vital in scientific applications. Adaptive mesh refinement (AMR) is a key technique for reducing computational cost by concentrating resolution in regions of interest. In this work, we develop an AMR-based numerical solver using Regent, a high-level programming language for the Legion programming model. We address several challenges associated with implementing AMR in Regent. These include dynamic data structures for patch refinement/coarsening, mesh validity enforcement, and reducing task launch overhead via task fusion. Experimental results show that task fusion achieves 18x speedup, while automated GPU kernel generation via simple annotations yields 9.7x speedup for the targeted kernel. We demonstrate our approach through simulations of two canonical compressible flow problems governed by the Euler equations.

Problem

• Implementing AMR in Regent for compressible flow simulations
• Addressing dynamic data structures for patch refinement/coarsening
• Reducing task launch overhead via task fusion

Innovation

• Task-based programming with Regent language
• Dynamic data structures for AMR patches
• Task fusion and GPU kernel generation