The Processing-in-Memory Model

The Processing-in-Memory Model
复制标题

内存处理模型

DOI:
10.1145/3409964.3461816
复制
发表时间:
2021
期刊:
SPAA '21: 33rd ACM Symposium on Parallelism in Algorithms and Architectures
影响因子:
--
通讯作者:
McGuffey, Charles
McGuffey, Charles
中科院分区:
--
文献类型:
--
作者:
Kang, Hongbo;Gibbons, Phillip B.;Blelloch, Guy E.;Dhulipala, Laxman;Gu, Yan;McGuffey, Charles

文献摘要

参考文献

被引文献

相似文献

随着计算资源变得更加高效和数据规模不断增长,数据移动正迅速成为计算中的主要成本。内存处理正在成为减少昂贵的数据移动的关键技术,它使计算能够在嵌入内存模块本身的计算资源上执行。本文提出了内存处理 (PIM) 模型,用于在提供内存处理模块的系统上设计和分析并行算法。 PIM 模型侧重于此类系统的关键方面,同时抽象了其余部分。也就是说,该模型结合了 (i) 由并行核心组成的 CPU 端,可快速访问大小为 M 个字的小型共享内存(如传统并行计算中),(ii) 由 P 个 PIM 模块组成的 PIM 端,每个模块都有一个核心和大小为 θ(n/P) 个字的本地内存,用于大小为 n 的输入(如在传统分布式计算中),以及 (iii) 两侧之间的网络。该模型结合了共享内存(工作和深度)和分布式内存(本地工作、通信时间)计算的标准并行复杂度指标。一个关键的算法挑战是在 PIM 模块之间的通信和本地工作中实现负载平衡,同时最大限度地减少通信时间。我们演示了如何克服有序搜索结构的这一挑战,提出了一种并行 PIM-skiplist 数据结构,该结构有效地支持各种批量并行查询和更新。
As computational resources become more efficient and data sizes grow, data movement is fast becoming the dominant cost in computing. Processing-in-Memory is emerging as a key technique for reducing costly data movement, by enabling computation to be executed on compute resources embedded in the memory modules themselves.This paper presents the Processing-in-Memory (PIM) model, for the design and analysis of parallel algorithms on systems providing processing-in-memory modules. The PIM model focuses on keys aspects of such systems, while abstracting the rest. Namely, the model combines (i) a CPU-side consisting of parallel cores with fast access to a small shared memory of size M words (as in traditional parallel computing), (ii) a PIM-side consisting of P PIM modules, each with a core and a local memory of size Θ(n/P) words for an input of size n (as in traditional distributed computing), and (iii) a network between the two sides. The model combines standard parallel complexity metrics for both shared memory (work and depth) and distributed memory (local work, communication time) computing. A key algorithmic challenge is to achieve load balance among the PIM modules in both their communication and their local work, while minimizing the communication time. We demonstrate how to overcome this challenge for an ordered search structure, presenting a parallel PIM-skiplist data structure that efficiently supports a wide range of batch-parallel queries and updates.
DOI: 10.1201/9781439833032.ch197
发表时间: 2007-05
期刊: --
影响因子: --
作者:
Kevin Henry
通讯作者: Kevin Henry
只需加入并行有序集
DOI: 10.1145/2935764.2935768
发表时间: 2016
期刊: Proceedings of the 28th ACM Symposium on Parallelism in Algorithms and Architectures
影响因子: --
作者:
G. Blelloch;Daniel Ferizovic;Yihan Sun
通讯作者: Yihan Sun
DOI: 10.1016/j.tcs.2008.09.023
发表时间: 2005-02
期刊: Theor. Comput. Sci.
影响因子: --
作者:
P. Berenbrink;T. Friedetzky;Zengjian Hu;R. Martin
通讯作者: P. Berenbrink;T. Friedetzky;Zengjian Hu;R. Martin
DOI: 10.1145/3323165.3323191
发表时间: 2019-06
期刊: The 31st ACM Symposium on Parallelism in Algorithms and Architectures
影响因子: --
作者:
Jiwon Choe;Amy Huang;T. Moreshet;Maurice Herlihy;R. I. Bahar
通讯作者: Jiwon Choe;Amy Huang;T. Moreshet;Maurice Herlihy;R. I. Bahar
顺序随机排列、列表收缩和树收缩是高度并行的
DOI: 10.1137/1.9781611973730.30
发表时间: 2015
期刊: Commun. ACM
影响因子: --
作者:
Julian Shun;Yan Gu;G. Blelloch;Jeremy T. Fineman;Phillip B. Gibbons
通讯作者: Phillip B. Gibbons