标题： 剖析AMD MI300A APU上的CPU-GPU统一物理内存 显示英文标题

标题： Dissecting CPU-GPU Unified Physical Memory on AMD MI300A APUs

摘要： 离散GPU是HPC和数据中心系统的核心，需要管理独立的CPU和GPU内存空间。统一虚拟内存（UVM）被提出以减轻内存管理的负担；然而，这会带来性能上的高昂代价。最近，AMD推出的MI300A加速处理单元（APU）——在El Capitan超级计算机中部署——首次实现了集成CPU和GPU的HPC系统，具备统一物理内存（UPM）。本工作首次对MI300A上的UPM架构进行了全面表征。我们首先分析了UPM系统特性，包括内存延迟、带宽和一致性开销。然后评估了系统软件在内存分配、页面故障处理、TLB管理和Infinity Cache使用方面的效率。我们提出了将应用程序转换为UPM架构的一组移植策略，并在MI300A APU上评估了六个应用程序。我们的结果表明，使用统一内存模型的UPM应用程序可以与显式管理模型的应用程序相媲美或表现更优——同时将内存成本降低了高达44%。 显示英文摘要

摘要： Discrete GPUs are a cornerstone of HPC and data center systems, requiring management of separate CPU and GPU memory spaces. Unified Virtual Memory (UVM) has been proposed to ease the burden of memory management; however, at a high cost in performance. The recent introduction of AMD's MI300A Accelerated Processing Units (APUs)--as deployed in the El Capitan supercomputer--enables HPC systems featuring integrated CPU and GPU with Unified Physical Memory (UPM) for the first time. This work presents the first comprehensive characterization of the UPM architecture on MI300A. We first analyze the UPM system properties, including memory latency, bandwidth, and coherence overhead. We then assess the efficiency of the system software in memory allocation, page fault handling, TLB management, and Infinity Cache utilization. We propose a set of porting strategies for transforming applications for the UPM architecture and evaluate six applications on the MI300A APU. Our results show that applications on UPM using the unified memory model can match or outperform those in the explicitly managed model--while reducing memory costs by up to 44%.