arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.06204cs.AR
收藏

PIM-MMU:一种用于加速商用 PIM 系统中数据传输的内存管理单元

PIM-MMU: A Memory Management Unit for Accelerating Data Transfers in Commercial PIM Systems

Dongjae Lee, Bongjoon Hyun, Taehun Kim, Minsoo Rhu

AI总结:

针对 UPMEM 商用 PIM 系统中 DRAM-PIM 数据传输瓶颈,提出软硬件协同的 PIM-MMU,将传输吞吐与能效均提升 4.1 倍,并实现 2.2 倍端到端加速。

摘要:

内存内处理(PIM)已成为加速内存密集型工作负载的一种有前景的解决方案,因为它能为处理单元提供高内存带宽。这种方法不仅引起了学术界的关注,也引起了工业界的关注,从而推动了真实商用 PIM 设备的开发。在本研究中,我们首先对 UPMEM 的通用 PIM 系统进行深入表征,并分析跨 DRAM 与 PIM 地址空间的数据传输所造成的瓶颈。我们的表征研究揭示了内存总线集成式 PIM 系统中 DRAM 与 PIM 之间数据传输相关的若干关键挑战,例如其较高的 CPU 核心利用率、较高的功耗,以及 DRAM 和 PIM 两者较低的读/写吞吐量。受这些关键发现驱动,我们提出 PIM-MMU 架构,这是一种硬件/软件协同设计,能够为 PIM 系统实现高能效的 DRAM 与 PIM 之间传输。PIM-MMU 协同结合了基于硬件的数据复制引擎、面向 PIM 优化的内存调度器以及异构感知内存映射函数,并由我们的 PIM-MMU 软件栈支持其使用,从而显著提高 DRAM 与 PIM 之间数据传输的效率。实验结果表明,PIM-MMU 将 DRAM 与 PIM 之间的数据传输吞吐量平均提高 4.1 倍,并将其能效提高 4.1 倍,从而为真实 PIM 工作负载带来 2.2 倍的端到端加速。

英文摘要:

Processing-in-memory (PIM) has emerged as a promising solution for accelerating memory-intensive workloads as they provide high memory bandwidth to the processing units. This approach has drawn attention not only from the academic community but also from the industry, leading to the development of real-world commercial PIM devices. In this work, we first conduct an in-depth characterization on UPMEM's general purpose PIM system and analyze the bottlenecks caused by the data transfers across the DRAM and PIM address space. Our characterization study reveals several critical challenges associated with DRAM to/from PIM data transfers in memory bus integrated PIM systems, for instance, its high CPU core utilization, high power consumption, and low read/write throughput for both DRAM and PIM. Driven by our key findings, we introduce the PIM-MMU architecture which is a hardware/software codesign that enables energy-efficient DRAM to/from PIM transfers for PIM systems. PIM-MMU synergistically combines a hardwarebased data copy engine, a PIM-optimized memory scheduler, and a heterogeneity-aware memory mapping function, the utilization of which is supported by our PIM-MMU software stack, significantly improving the efficiency of DRAM to/from PIM data transfers. Experimental results show that PIM-MMU improves the DRAM to/from PIM data transfer throughput by an average 4.1x and enhances its energy-efficiency by 4.1x, leading to a 2.2x end-to-end speedup for real-world PIM workloads.

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。

↑