PolyGraph: Exposing the Value of Flexibility for Graph Processing Accelerators
PolyGraph: Exposing the Value of Flexibility for Graph Processing Accelerators
复制标题
DOI：
10.1109/isca52012.2021.00053
复制
发表时间：
2021-06
期刊：
影响因子：
--
通讯作者：
Vidushi Dadu;Sihao Liu;Tony Nowatzki
中科院分区：
文献类型：
--
作者：
Vidushi Dadu;Sihao Liu;Tony Nowatzki
由于图形工作负载的重要性和CPU/GPU的局限性，人们提出了许多图形处理加速器。现有加速器的基本方法是专注于单个图形算法变体(例如，批量同步+切片)。虽然这对专门化很有帮助，但这使表上的灵活性失去了性能潜力，并使理解图形类型、工作负载、算法和专门化之间的关系变得复杂。在这项工作中，我们探索了灵活性在图形处理加速器中的价值。首先，我们确定了关键算法变体的分类。然后我们开发了一个模板架构(Polygraph)，它可以灵活地跨越这些变体，同时能够以模块方式集成每个变体的专门化功能。总体而言，我们发现图形加速的灵活性是关键。如果只支持一种变体，asynchronous-updates/priority-vertex-scheduling/graph-slicing是最好的设计，与性能最好的加速器GraphPulse相比，它的加速比可达到1.93倍。但是，每工作负载的静态灵活性可以将性能进一步提高2.71倍。凭借每阶段的动态灵活性，性能进一步提升高达50%。
Because of the importance of graph workloads and the limitations of CPUs/GPUs, many graph processing accelerators have been proposed. The basic approach of prior accelerators is to focus on a single graph algorithm variant (eg. bulk-synchronous + slicing). While helpful for specialization, this leaves performance potential from flexibility on the table and also complicates understanding the relationship between graph types, workloads, algorithms, and specialization.In this work, we explore the value of flexibility in graph processing accelerators. First, we identify a taxonomy of key algorithm variants. Then we develop a template architecture (PolyGraph) that is flexible across these variants while being able to modularly integrate specialization features for each.Overall we find that flexibility in graph acceleration is critical. If only one variant can be supported, asynchronous-updates/priority-vertex-scheduling/graph-slicing is the best design, achieving 1.93× speedup over the best-performing accelerator, GraphPulse. However, static flexibility per-workload can further improve performance by 2.71×. With dynamic flexibility per-phase, performance further improves by up to 50%.