SPX: Collaborative Research: Global Address Programming with Accelerators

SPX：协作研究：使用加速器进行全局地址编程

基本信息

批准号：
1823037
负责人：
John Owens
金额：
$ 38.6万
依托单位：
University of California-Davis
依托单位国家：
美国
项目类别：
Standard Grant
财政年份：
2018
资助国家：
美国
起止时间：
2018-10-01 至 2022-09-30
项目状态：
已结题

来源：
https://www.nsf.gov/awardsearch/showAward?AWD_ID=1823037&HistoricalAwards=false
关键词：
SPX Collaborative Research Global Address

项目摘要

Large-scale computing today is dominated by parallel computing, where a large task is divided into many smaller tasks and those smaller tasks run at the same time. Traditionally each of those tasks run independently up to a common stopping point, then they halt, exchange information, and continue. This global stop-and-communicate step is quite expensive. This project instead pursues a different approach, where individual tasks directly communicate with other tasks asynchronously, without having to wait for a global stopping point. This approach is likely to yield better performance on large-scale computing tasks, specifically on what is becoming the dominant large-scale machine, a heterogeneous machine with many CPUs and othermany-core processors. The project will deliver a set of high-performance, open-source data structures and algorithm implementations to support irregular patterns of communication, notably those that arise in biology, graph analytics, and sparse linear algebra for machine learning. These will not only be directly useful for end users but also demonstrate how to design and engineer primitives for accelerator-equipped distributed-memory machines. The project also engages application developers (both in our groups and externally) to make the outcomes broadly useful.The project will develop a programming environment for accelerator-based HPC systems that integrates accelerators into a Partitioned Global Address Space (PGAS) model, which will allow direct communication between GPUs in a manner that is well suited to both applications and the underlying hardware. Specifically, GPU programming will be integrated with the UPC++ PGAS programming model ("GPUPC++"). The project will thus advance the state of the art in algorithms, programming models, and low-level support for the heterogeneous large-scale computers.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

如今的大规模计算以并行计算为主，其中一个大任务被划分为许多较小的任务，并且这些较小的任务同时运行。传统上，这些任务中的每一个都独立运行到一个共同的停止点，然后它们停止、交换信息并继续。这种全球性的停止和沟通步骤是相当昂贵的。相反，该项目采用了不同的方法，其中各个任务直接与其他任务异步通信，而无需等待全局停止点。这种方法可能会在大规模计算任务上产生更好的性能，特别是在正在成为主导的大型机器上，即具有许多 CPU 和其他多核处理器的异构机器上。该项目将提供一组高性能、开源数据结构和算法实现，以支持不规则的通信模式，特别是生物学、图形分析和机器学习的稀疏线性代数中出现的通信模式。这些不仅对最终用户直接有用，而且还演示如何为配备加速器的分布式内存机器设计和工程原语。该项目还吸引了应用程序开发人员（包括我们团队和外部的），以使成果具有广泛的用途。该项目将为基于加速器的 HPC 系统开发一个编程环境，将加速器集成到分区全局地址空间 (PGAS) 模型中，这将允许 GPU 之间以非常适合应用程序和底层硬件的方式进行直接通信。具体来说，GPU编程将与UPC++ PGAS编程模型（“GPUPC++”）集成。因此，该项目将推进算法、编程模型和对异构大型计算机的低级支持方面的最先进水平。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力优点和技术进行评估，被认为值得支持。更广泛的影响审查标准。

项目成果

期刊论文数量（5）

专著数量（0）

科研奖励数量（0）

会议论文数量（0）

专利数量（0）

A Roadmap for the GraphBLAS C++ API

GraphBLAS C API 路线图

DOI：
10.1109/ipdpsw50202.2020.00049
发表时间：
2020
期刊：
and Learning (IPDPSW
影响因子：
0
作者：
Brock, Benjamin;Buluc, Aydin;Mattson, Timothy G.;McMillan, Scott;Moreira, Jose E.
通讯作者：
Moreira, Jose E.

RDMA vs. RPC for Implementing Distributed Data Structures

DOI：
10.1109/ia349570.2019.00009
发表时间：
2019-10
期刊：
2019 IEEE/ACM 9th Workshop on Irregular Applications: Architectures and Algorithms (IA3)
影响因子：
0
作者：
Benjamin Brock;Yuxin Chen;Jiakun Yan;John Douglas Owens;A. Buluç;K. Yelick
通讯作者：
Benjamin Brock;Yuxin Chen;Jiakun Yan;John Douglas Owens;A. Buluç;K. Yelick

Considerations for a Distributed GraphBLAS API

分布式 GraphBLAS API 的注意事项

DOI：
10.1109/ipdpsw50202.2020.00048
发表时间：
2020
期刊：
and Learning
影响因子：
0
作者：
Brock, Benjamin;Buluc, Aydin;Mattson, Timothy G.;McMillan, Scott;Moreira, Jose E.;Pearce, Roger;Selvitopi, Oguz;Steil, Trevor
通讯作者：
Steil, Trevor

Atos: A Task-Parallel GPU Scheduler for Graph Analytics

Atos：用于图形分析的任务并行 GPU 调度程序

DOI：
10.1145/3545008.3545056
发表时间：
2022
期刊：
Proceedings of the 51st International Conference on Parallel Processing
影响因子：
0
作者：
Chen, Yuxin;Brock, Benjamin;Porumbescu, Serban;Buluc, Aydin;Yelick, Katherine;Owens, John
通讯作者：
Owens, John

Scalable Irregular Parallelism with GPUs: Getting CPUs Out of the Way

使用 GPU 实现可扩展的不规则并行：让 CPU 不再碍事

DOI：
发表时间：
2022
期刊：
International Conference for High Performance Computing Networking Storage and Analysis
影响因子：
0
作者：
Chen, Yuxin;Brock, Benjamin;Porumbescu, Șerban;Buluc, Aydın;Yelick, Katherine;Owens, John D.
通讯作者：
Owens, John D.