AI-coupled HPC Workflow Applications, Middleware and Performance

2024-06-20

arXiv.org

AI Summary

This study addresses critical challenges in the deep integration of AI and high-performance computing (HPC), including heterogeneous coupling patterns, rapid co-evolution, and ambiguous performance bottlenecks. Methodologically, it systematically classifies six canonical AI–HPC collaborative execution patterns; introduces the first dynamic conceptual framework for performance analysis; and proposes a general analytical paradigm grounded in cross-layer workflow modeling, coupling-pattern abstraction, bottleneck diagnosis, middleware requirements analysis, and benchmark suite design. The contributions include: (1) identification of multi-level performance barriers across software, runtime, and hardware stacks; (2) clear articulation of core challenges—such as adaptive orchestration, latency-sensitive data movement, and heterogeneity-aware scheduling—and their evolutionary boundaries; and (3) formulation of open scientific questions demanding urgent investigation. Furthermore, the work provides a methodological foundation and strategic guidance for developing domain-specific AI–HPC benchmarking frameworks, thereby advancing co-design principles for next-generation intelligent HPC systems.

Technology Category

• Humans and AI: Other Foundations of Human Computation & AI
• Machine Learning: Hardware-aware ML
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• Economics, Online Markets and Human Computation: Research challenges in human and human-AI computation
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs

Abstract

AI integration is revolutionizing the landscape of HPC simulations, enhancing the importance, use, and performance of AI-driven HPC workflows. This paper surveys the diverse and rapidly evolving field of AI-driven HPC and provides a common conceptual basis for understanding AI-driven HPC workflows. Specifically, we use insights from different modes of coupling AI into HPC workflows to propose six execution motifs most commonly found in scientific applications. The proposed set of execution motifs is by definition incomplete and evolving. However, they allow us to analyze the primary performance challenges underpinning AI-driven HPC workflows. We close with a listing of open challenges, research issues, and suggested areas of investigation including the the need for specific benchmarks that will help evaluate and improve the execution of AI-driven HPC workflows.

Problem

• Survey AI-driven HPC workflows and conceptual basis
• Propose six common execution motifs in scientific applications
• Analyze performance challenges in AI-driven HPC workflows

Innovation

• AI-coupled HPC workflow integration
• Six execution motifs for AI-HPC
• Performance benchmarks for AI-HPC