Optimization of asynchronous graph processing on GPU with hybrid coloring model

Shi, Xuanhua; Liang, Junling; Di, Sheng; He, Bingsheng; Jin, Hai; Lu, Lu; Wang, Zhixiang; Luo, Xuan; Zhong, Jianlong

doi:10.1145/2688500.2688542

Cited by 18 publications

(15 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In other words, the result of the offloaded task is not required by other tasks for their execution. Asynchronous execution split up the problem into multiple tasks and process them independently [26], [27]. A real-life example of asynchronous execution is discussion forums where every user can post their views independent of any other user.…”

Section: Terminologiesmentioning

confidence: 99%

Comparative Analysis of Network Libraries for Offloading Efficiency in Mobile Cloud Environment

Sufyan¹,

Banerjee²

2019

ijacsa

View full text Add to dashboard Cite

In the modern era, smartphones are increasingly becoming an integral and essential part of our daily life. Although the hardware capabilities of the smartphones (i.e., processing, memory, battery, and communication) are improving every day, however, it is not enough to handle computationintensive applications, such as image processing, data analytics, and encryption. To overcome these limitations, mobile cloud computing (MCC) is introduced, which augments the capabilities of smartphones and resources of the cloud to provide better QoS performance to the user. The idea is to save resources in the smartphones by offloading the computationally intensive tasks to the cloud. In this context, researchers have proposed several offloading frameworks, mainly addressing challenges of why-what-when and where to offload. In this paper, however, we explore another challenging issue of offloading, i.e., howto-offload. More specifically, we analyze different networking libraries (HttpURLConnection, OkHttp, Volley, Retrofit) and study their performance on various dynamic factors such as data size, communication media, hardware and software of the smartphone. Our objective is to explore if an application can use the same networking library for all the smartphones and all purposes or there is a need to make an adaptive decision based on the local constraints. To understand this, we perform a comprehensive analysis of the networking libraries on different Andriod smartphones in the real environment and found that there is a need of adaptive network library selection because libraries perform changes in different scenarios.

show abstract

Section: Terminologiesmentioning

confidence: 99%

Comparative Analysis of Network Libraries for Offloading Efficiency in Mobile Cloud Environment

Sufyan¹,

Banerjee²

2019

ijacsa

View full text Add to dashboard Cite

show abstract

“…GraphR [9] 也采用这种分割方式, 利用 ReRAM 的存储特性来设计高性能低功耗的图计算加速器. [24] , Medusa [25] , MapGraph [26] , Frog [27] 等都是前面提到的 CSR 的布局方式, 这种存储模式可以压缩图数据占据的内存空间大小. 但是使用 CSR 会在运算过程中对顶点属性的访问产生很大的随机性, 造成了大量的随机访问.…”

Section: 图计算常见编程模型unclassified

“…[Gemini [16] , CPU, 2016], [Frog [27] , GPU, 2015], [Page [46] , CPU, 2015], [GraphReduce [31] , GPU, 2015], etc.…”

Section: 新兴存储器技术利用新兴存储器技术将计算逻辑集成在存储器内是一个很好的降低能耗的方式例如前所述的 Hmcmentioning

confidence: 99%

“…[REBFS [36] , PIM, 2018], [Tesseract [37] , PIM, 2015], [GraphH [19] , PIM, 2018], [Graphicionado [2] , ASIC, 2016], [GraphGen [5] , FPGA, 2014], [FPGA-HMC [38] , FPGA, 2018], [GTS [29] , GPU, 2016], [Frog [27] , GPU, 2015], etc.…”

Section: 新兴存储器技术利用新兴存储器技术将计算逻辑集成在存储器内是一个很好的降低能耗的方式例如前所述的 Hmcmentioning

confidence: 99%

See 1 more Smart Citation

Memory system optimization for graph processing: a survey

Wang¹,

Zhang²,

Wang³

et al. 2019

Sci. Sin.-Inf.

View full text Add to dashboard Cite

现有的工作针对不同的架构和平台进行了多方面尝试, 主要的架构平台包括高性能 GPU 架构、基于 PIM (processing in memory) 的架构以及专用图计算加速器 (ASIC 和 FPGA). GPU 的高并行度与高带宽对图计算的运算加速起到了至关重要的作用; 为了减少内存的访问带来的开销, 人们设计了各种数据格式和图算法以提高访存效率. PIM 技术让许多新兴硬件应用到图计算中, 如 HMC (hybrid memory cube), ReRAM (resistive RAM) 等, 实现访存一体化, 减少内存数据的移动传输带来的开销. 突破通用处理器的限制, 图计算专用加速器包括基于 FPGA 的设计以及专用芯片 (ASIC) 的设计, 尝试自主设计运算流水线、访存模式等优化策略来加速图计算运算效率以及减少能耗. 本文针对图计算加速环境下不同架构平台下对于内存管理所作出的努力以及其他方面的优化方案进行综述. 2 图计算基本介绍本节首先介绍图计算的基本概念与图算法的特点分类, 然后总结图计算的计算特征和常见的编程模型. 2.1 图的基本定义图是一种由顶点 (vertices) 和连接顶点的边 (edges) 构成的数据结构, G = (V, E), 其中 V 表示顶点集合, E 表示边的集合. e = (v i , v j ) 表示从顶点 v i 到 v j 的一条有向边. 同时, 每一个顶点与每一条边都有属于自己的属性值. 不同的领域属性值可代表不同的含义. 如社交网络中, 顶点的属性值为个人的热度, 而边的属性值则表示为有关联的两人之间的紧密程度. 现实中的自然图存在以下的特性.(1) 稀疏性. 顶点的平均度数很小, 数据分布比较分散, 图数据的稀疏性会导致较差的局部性数据访问和大量的随机访问.(2) 幂律性. 幂律性是自然图的一个很常见的特征, 图数据中一小部分顶点关联着绝大部分的边. 而大部分的顶点只关联着少量的边. 图数据的幂律性会导致很严重的负载不均衡.

show abstract

“…Previous work indicated that asynchronous approaches have the potential to minimize the effects of load imbalance across different cores in multicore machines [18]. GraphLab [19], PowerGraph [20], and other distributed frameworks [21,22] have investigated the use of asynchronous execution models, which could implement the traversal operations in general. However, these approaches are more suitable for the distributed, batch-oriented graph computation that runs on the entire graph, instead of interactive traversal and query of subgraphs, which are common in our HPC-rich metadata management system.…”

Section: Introductionmentioning

confidence: 99%

An asynchronous traversal engine for graph-based rich metadata management

et al. 2016

View full text Add to dashboard Cite

Rich metadata in high-performance computing (HPC) systems contains extended information about users, jobs, data files, and their relationships. Property graphs are a promising data model to represent heterogeneous rich metadata flexibly. Specifically, a property graph can use vertices to represent different entities and edges to record the relationships between vertices with unique annotations. The high-volume HPC use case, with millions of entities and relationships, naturally requires an out-of-core distributed property graph database, which must support live updates (to ingest production information in real time), low-latency point queries (for frequent metadata operations such as permission checking), and large-scale traversals (for provenance data mining). Among these needs, large-scale property graph traversals are particularly challenging for distributed graph storage systems. Most existing graph systems implement a "level-synchronous" breadth-first search algorithm that relies on global synchronization in each traversal step. This performs well in many problem domains; but a rich metadata management system is characterized by imbalanced graphs, long traversal lengths, and concurrent workloads, each of which has the potential to introduce or exacerbate stragglers (i.e., abnormally slow steps or servers in a graph traversal) that lead to low overall throughput for synchronous traversal algorithms. Previous research indicated that the straggler problem can be mitigated by using asynchronous traversal algorithms, and many graph-processing frameworks have successfully demonstrated this approach. Such systems require the graph to be loaded into a separate batch-processing framework instead of being iteratively accessed, however. In this work, we investigate a general asynchronous graph traversal engine that can operate atop a rich metadata graph in its native format. We outline a traversal-aware query language and key optimizations (traversal-affiliate caching and execution merging) necessary for efficient performance. We further explore the effect of different graph partitioning strategies on the traversal performance for both synchronous and asynchronous traversal engines. Our experiments show that the asynchronous graph traversal engine is more efficient than its synchronous counterpart in the case of HPC rich metadata processing, where more servers are involved and larger traversals are needed. Moreover, the asynchronous traversal engine is more adaptive to different graph partitioning strategies.

show abstract

Optimization of asynchronous graph processing on GPU with hybrid coloring model

Cited by 18 publications

References 5 publications

Comparative Analysis of Network Libraries for Offloading Efficiency in Mobile Cloud Environment

Comparative Analysis of Network Libraries for Offloading Efficiency in Mobile Cloud Environment

Memory system optimization for graph processing: a survey

An asynchronous traversal engine for graph-based rich metadata management

Contact Info

Product

Resources

About