How Well do CPU, GPU and Hybrid Graph Processing Frameworks Perform?

Aasawat, Tanuj Kr; Reza, Tahsin; Ripeanu, Matei

doi:10.1109/ipdpsw.2018.00082

Cited by 4 publications

(4 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Further, the largest graph submission we made has 128 Billion undirected edges (RMAT32), with edge list size 1TB. NUMA-oblivious Totem could not run RMAT32 as its memory requirement are ∼2× the edge-list size, as observed in our previous study [5].…”

Section: Graph500 Submissionsmentioning

confidence: 79%

“…We consider PageRank, Breadth-First Search -Top Down (BFS-TD), Breadth-First Search -Direction Optimized (BFS-DO), and Single-Source Shortest Path (SSSP) algorithms. We use these algorithms to evaluate our work because (i) these algorithms have been widely studied in the context of high-performance graph processing systems and have been used in the past studies [5,15,17,30,35,36,42], (ii) BFS and SSSP are also used as benchmarks for the Graph500 competition [2], to rank supercomputers for data intensive applications, (iii) they are the building blocks of more complex graph algorithms (for example, BFS is used as a subroutine in complex graph algorithms like connected components, max flow, betweeness centrality and clustering), and (iv) these algorithms are good representation for studying the performance of any hardware platform for irregular memory access pattern.…”

Section: Graph Algorithmsmentioning

confidence: 99%

“…And since NUMA architecture is also shared-memory based, these frameworks run on NUMA architecture as well. But, they suffer from the distributed nature of the shared-memory in NUMA, and hence do not perform and scale well [5]. NUMA-aware Graph Processing Framework.…”

Section: Related Workmentioning

confidence: 99%

“…Similar to distributed systems, it follows Bulk Synchronous Parallel processing model, and does communication between CPU and GPU with message buffers. We use this NUMAoblivious framework because: First, in our previous study [5] we observed that it outperforms state-of-the-art graph processing frameworks including Intel's Graph-Mat [36] and Galois [30], by up to an order of magnitude. Second, and most importantly, its processing model, BSP, matches with our needs.…”

Section: System Implementationmentioning

confidence: 99%

See 3 more Smart Citations

Scale-Free Graph Processing on a NUMA Machine

Aasawat¹,

Reza²,

Ripeanu³

2018

2018 IEEE/ACM 8th Workshop on Irregular Applications: Architectures and Algorithms (IA3)

Self Cite

View full text Add to dashboard Cite

The importance of high-performance graph processing to solve big data problems targeting high-impact applications is greater than ever before. Graphs incur highly irregular memory accesses which leads to poor data locality, load imbalance, and data-dependent parallelism. Distributed graph processing frameworks, such as Google's Pregel, that employs memory-parallel, shared-nothing systems have experienced tremendous success in terms of scale and performance. Modern sharedmemory systems embrace the so called Non-Uniform Memory Access (NUMA) architecture which has proven to be more scalable (in terms of numbers of cores and memory modules) than the Symmetric Multiprocessing (SMP) architecture. In many ways, a NUMA system resembles a shared-nothing distributed system: physically distinct processing cores and memory regions (although, cache-coherent in NUMA). Memory accesses to remote NUMA domains are more expensive than local accesses. This poses the opportunity to transfer the know-how and design of distributed graph processing to develop shared-memory graph processing solutions optimized for NUMA systems (which is surprisingly little-explored).In this dissertation, we explore if a distributed-memory like middleware that makes graph partitioning and communication between partitions explicit, can improve the performance on a NUMA system. We design and implement a NUMA aware graph processing framework that treats the NUMA platform as a distributed system, and embraces its design principles; in particular explicit partitioning and inter-partition communication. We further explore design trade-offs to reduce communication overhead and propose a solution that embraces design philosophies of distributed graph processing system and at the same time exploits optimization opportunities specific to single-node systems. We demonstrate up to 13.9× speedup iii over a state-of-the-art NUMA-aware framework, Polymer and up to 3.7× scalability on a four-socket machine using graphs with tens of billions of edges.iv Preface This thesis is based on the research project done by me under the supervision and guidance of Professor Matei Ripeanu. I was responsible for the design, implementation, modeling, validation, evaluation and analysis of the results, along with taking the lead in publication writing effort. The research presented in this thesis have been either published or accepted for publication.The work that this thesis extends and evaluates against, was selected based on the following preliminary study; Professor Ripeanu and Tahsin helped me in the analysis of the results and editing the publication.

show abstract

Section: Graph500 Submissionsmentioning

confidence: 79%