DiOMP-Offloading：面向可移植分布式异构OpenMP
DiOMP-Offloading: Toward Portable Distributed Heterogeneous OpenMP
AI总结：
针对传统HPC混合编程模型在分布式GPU内存管理与可移植性上的不足，提出分布式OpenMP框架DiOMP，融合OpenMP目标卸载与PGAS模型，实验证明其在多类异构平台上具备更优性能、可扩展性与可编程性。
摘要：
随着高性能计算（HPC）中核心数量与异构性不断提升，传统混合编程模型在管理分布式GPU内存、保障可移植性方面面临挑战。本文提出DiOMP，这是一款分布式OpenMP框架，将OpenMP目标卸载与分区全局地址空间（PGAS）模型相统一。该框架基于LLVM/OpenMP构建，采用GASNet-EX或GPI-2实现通信，可透明处理全局内存，同时支持对称与非对称GPU内存分配，还利用了兼容厂商库的可移植集合通信层OMPCCL。DiOMP通过抽象设备内存与通信简化编程，相比传统方案具备更优的可扩展性与可编程性。在NVIDIA A100、Grace Hopper、AMD MI250X平台上的评估显示，其在微基准测试以及矩阵乘法、Minimod等应用中性能均有提升，展现了DiOMP在可扩展、可移植、高效异构计算领域的潜力。
英文摘要：
As core counts and heterogeneity rise in HPC, traditional hybrid programming models face challenges in managing distributed GPU memory and ensuring portability. This paper presents DiOMP, a distributed OpenMP framework that unifies OpenMP target offloading with the Partitioned Global Address Space (PGAS) model. Built atop LLVM/OpenMP and using GASNet-EX or GPI-2 for communication, DiOMP transparently handles global memory, supporting both symmetric and asymmetric GPU allocations. It leverages OMPCCL, a portable collective communication layer compatible with vendor libraries. DiOMP simplifies programming by abstracting device memory and communication, achieving superior scalability and programmability over traditional approaches. Evaluations on NVIDIA A100, Grace Hopper, and AMD MI250X show improved performance in micro-benchmarks and applications like matrix multiplication and Minimod, highlighting DiOMP's potential for scalable, portable, and efficient heterogeneous computing.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
