SC13 Research Highlight: Large Graph Processing Without the Overhead

By Dr. Ling Liu and Kisung Lee

November 16, 2013

Many real world information networks consist of millions or billions of vertices representing heterogeneous entities and billions or trillions of edges representing heterogeneous types of relationships among entities.

Image Source: Max Delbrück Center for Molecular Medicine

For example, the crawled Web graph is estimated to have more than 20 billions of pages (vertices) with 160 billions hyperlinks (edges). Facebook user community exceeds 1 billion users (vertices) with more than 140 billion friendship relationships (edges) in 2012.  The billion triple challenges from the Semantic Web community have put forward large collection of RDF datasets with hundreds of millions of vertices and billions of edges.

As the size and variety of information networks continue to grow in many science and engineering domains, graph computations often exceed the processing capacity of conventional hardware, software systems and tools for a number of reasons. First, graph data often exhibits higher data correlations through both direct and indirect edges and such high correlation tends to generate large size of intermediate results during graph computations. When the size of intermediate results exceeds the available memory, the out of memory problem is unavoidable. Second, the graph datasets are growing in volume, variety and velocity. The bigger the size of the graphs gets, the worse the performance will be for most of the graph computations. One open challenge in this space is how to effectively partition a large graph to enable efficient parallel processing of complex graph operations.

One of the papers to be presented at the ACM/IEEE SC13 conference, titled “Efficient data partitioning model for heterogeneous graphs in the Cloud”, proposes a flexible graph partitioning framework, called VB-partitioner. This work is co-authored by the doctorate student Kisung Lee and Prof. Dr. Ling Liu from the school of Computer Science at Georgia Institute of Technology. To make parallel graph computations highly efficient, an important design goal of VB-Partitioner is to devise graph partitioning techniques that can effectively minimize the inter-partition communication overhead and maximize the intra-partition computation capacity (local processing).

Concretely, the first prototype of the VB-Partitioner focuses on efficient processing of graph queries, namely finding all the subgraphs matching a given subgraph pattern. VB-Partitioner partitions a large graph in three steps.

  • First, it constructs three types of Vertex Blocks (in-VBs, out-VBs and bi-VBs) to capture the general graph processing locality.  Each vertex block has an anchor vertex.
  • Second, it constructs three types of k-hop Extended Vertex Blocks (in-EVBs, out-EVBs and bi-EVBs) to distribute vertex blocks with better query locality. Each EVB has one anchor vertex. It achieves query locality by employing controlled edge replication. The setting of k is determined by the radius of frequent query graphs in order to ensure that most frequently requested queries can be processed in parallel at all partitions with minimized inter-partition communication overhead.
  • Third, it partitions a graph by grouping its vertex blocks and EVBs to maximize parallelism in graph processing while ensuring load balance, controlled edge replication and fast grouping.

Four techniques are considered and compared in the context of grouping and placement of VBs and EVBs to partitions: random grouping, hash-based grouping, min-cut based grouping and high degree vertex-based grouping.  As an integral part of the VB-Partitioner, a data partition-aware query partitioning model is also developed to handle the cases where the radius of a query is larger than k. The experimental results reported in the paper demonstrate the effectiveness of VB-Partitioner in terms of query processing efficiency, data loading time and partition distribution balance.

Graph computations can be broadly classified into two categories, graph queries that find matching subgraphs of a given pattern and iterative graph algorithms that find clusters, orderings, paths or correlation patterns. The former targets at subgraph matching problems over large static graphs and the later targets at graph inference kernels that traverse the graph by iteratively updating the weight of vertices or edges, such as PageRank, shortest path algorithms, spanning tree algorithms, topological sort, and so forth.

Although the first generation of the VB-Partitioner is tailored primarily for efficient parallel processing of graph queries, the ongoing work on VB-Partitioner includes exploring the feasibility and effectiveness of VB-Partitioner in the context of iterative graph algorithms. For example, to minimize inter-partition communications and maximize parallelism in graph computation, it is crucial to optimize the shared memory by minimizing parallel overhead of synchronization barriers. It is equally important to optimize the distributed memory by bounding message buffer sizes, bundling messages, overlapping communication with computation to amortize the overhead of barriers.

image1
Image Source: Giot et al., “A Protein Interaction Map of Drosophila melanogaster”, Science 302, 1722-1736, 2003.”

In addition to exploring parallel computation opportunities through graph partitioning using multi-threads, multi-cores, multiple workers, one can also exploit and combine with other performance optimization techniques to scale large graph analytics. Example techniques include

  • Compression to provide compact storage and in-memory processing,
  • Data placements on disk and in memory to balance computation with storage, and to maximize sequential access and minimize random access,
  • Indexing at vertex and/or edge level to utilize sequential access and minimize unnecessary random access,
  • Caching at vertex, edge or query level to gain performance for repeated access.

Please come hear more on Tuesday, November 19, 2013 10:30AM – 11:00AM (Location: Room 205/207)

http://sc13.supercomputing.org/schedule/event_detail.php?evid=pap708

About the Authors

LingLing Liu is a Professor in the School of Computer Science at Georgia Institute of Technology. She directs the research programs in Distributed Data Intensive Systems Lab (DiSL), examining various aspects of large scale data intensive systems. Prof. Ling Liu is an internationally recognized expert in the areas of Cloud computing, Distributed Computing, Big Data technologies, Database systems and Service oriented computing. Prof. Liu is a recipient of IEEE Computer Society Technical Achievement Award in 2012. Currently Prof. Liu is the editor in chief of IEEE Transactions on Service Computing, and serves on the editorial board of half dozen international journals, including Journal of Parallel and Distributed Computing (JPDC), ACM Transactions on Internet Technology (TOIT), ACM Transactions on Web (TWEB). Dr. Liu’s current research is primarily sponsored by NSF, IBM, and Intel.

 

luiKisung Lee is a Ph.D student in the School of Computer Science at Georgia Tech. He received his BS and MS degree in computer science from KAIST in 2005 and 2007 respectively. He had worked for ETRI as a researcher from 2007 to 2010. He is conducting research in distributed and parallel processing of big data in the Cloud, mobile computing and social network analysis.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industy updates delivered to you every week!

Data Vortex Users Contemplate the Future of Supercomputing

October 19, 2017

Last month (Sept. 11-12), HPC networking company Data Vortex held its inaugural users group at Pacific Northwest National Laboratory (PNNL) bringing together about 30 participants from industry, government and academia t Read more…

By Tiffany Trader

AI Self-Training Goes Forward at Google DeepMind

October 19, 2017

DeepMind, Google’s AI research organization, announced today in a blog that AlphaGo Zero, the latest evolution of AlphaGo (the first computer program to defeat a Go world champion) trained itself within three days to play Go at a superhuman level (i.e., better than any human) – and to beat the old version of AlphaGo – without leveraging human expertise, data or training. Read more…

By Doug Black

Researchers Scale COSMO Climate Code to 4888 GPUs on Piz Daint

October 17, 2017

Effective global climate simulation, sorely needed to anticipate and cope with global warming, has long been computationally challenging. Two of the major obstacles are the needed resolution and prolonged time to compute Read more…

By John Russell

HPE Extreme Performance Solutions

Transforming Genomic Analytics with HPC-Accelerated Insights

Advancements in the field of genomics are revolutionizing our understanding of human biology, rapidly accelerating the discovery and treatment of genetic diseases, and dramatically improving human health. Read more…

Student Cluster Competition Coverage New Home

October 16, 2017

Hello computer sports fans! This is the first of many (many!) articles covering the world-wide phenomenon of Student Cluster Competitions. Finally, the Student Cluster Competition coverage has come to its natural home: H Read more…

By Dan Olds

Data Vortex Users Contemplate the Future of Supercomputing

October 19, 2017

Last month (Sept. 11-12), HPC networking company Data Vortex held its inaugural users group at Pacific Northwest National Laboratory (PNNL) bringing together ab Read more…

By Tiffany Trader

AI Self-Training Goes Forward at Google DeepMind

October 19, 2017

DeepMind, Google’s AI research organization, announced today in a blog that AlphaGo Zero, the latest evolution of AlphaGo (the first computer program to defeat a Go world champion) trained itself within three days to play Go at a superhuman level (i.e., better than any human) – and to beat the old version of AlphaGo – without leveraging human expertise, data or training. Read more…

By Doug Black

Student Cluster Competition Coverage New Home

October 16, 2017

Hello computer sports fans! This is the first of many (many!) articles covering the world-wide phenomenon of Student Cluster Competitions. Finally, the Student Read more…

By Dan Olds

Intel Delivers 17-Qubit Quantum Chip to European Research Partner

October 10, 2017

On Tuesday, Intel delivered a 17-qubit superconducting test chip to research partner QuTech, the quantum research institute of Delft University of Technology (TU Delft) in the Netherlands. The announcement marks a major milestone in the 10-year, $50-million collaborative relationship with TU Delft and TNO, the Dutch Organization for Applied Research, to accelerate advancements in quantum computing. Read more…

By Tiffany Trader

Fujitsu Tapped to Build 37-Petaflops ABCI System for AIST

October 10, 2017

Fujitsu announced today it will build the long-planned AI Bridging Cloud Infrastructure (ABCI) which is set to become the fastest supercomputer system in Japan Read more…

By John Russell

HPC Chips – A Veritable Smorgasbord?

October 10, 2017

For the first time since AMD's ill-fated launch of Bulldozer the answer to the question, 'Which CPU will be in my next HPC system?' doesn't have to be 'Whichever variety of Intel Xeon E5 they are selling when we procure'. Read more…

By Dairsie Latimer

Delays, Smoke, Records & Markets – A Candid Conversation with Cray CEO Peter Ungaro

October 5, 2017

Earlier this month, Tom Tabor, publisher of HPCwire and I had a very personal conversation with Cray CEO Peter Ungaro. Cray has been on something of a Cinderell Read more…

By Tiffany Trader & Tom Tabor

Intel Debuts Programmable Acceleration Card

October 5, 2017

With a view toward supporting complex, data-intensive applications, such as AI inference, video streaming analytics, database acceleration and genomics, Intel i Read more…

By Doug Black

Reinders: “AVX-512 May Be a Hidden Gem” in Intel Xeon Scalable Processors

June 29, 2017

Imagine if we could use vector processing on something other than just floating point problems.  Today, GPUs and CPUs work tirelessly to accelerate algorithms Read more…

By James Reinders

NERSC Scales Scientific Deep Learning to 15 Petaflops

August 28, 2017

A collaborative effort between Intel, NERSC and Stanford has delivered the first 15-petaflops deep learning software running on HPC platforms and is, according Read more…

By Rob Farber

Oracle Layoffs Reportedly Hit SPARC and Solaris Hard

September 7, 2017

Oracle’s latest layoffs have many wondering if this is the end of the line for the SPARC processor and Solaris OS development. As reported by multiple sources Read more…

By John Russell

US Coalesces Plans for First Exascale Supercomputer: Aurora in 2021

September 27, 2017

At the Advanced Scientific Computing Advisory Committee (ASCAC) meeting, in Arlington, Va., yesterday (Sept. 26), it was revealed that the "Aurora" supercompute Read more…

By Tiffany Trader

How ‘Knights Mill’ Gets Its Deep Learning Flops

June 22, 2017

Intel, the subject of much speculation regarding the delayed, rewritten or potentially canceled “Aurora” contract (the Argonne Lab part of the CORAL “ Read more…

By Tiffany Trader

Google Releases Deeplearn.js to Further Democratize Machine Learning

August 17, 2017

Spreading the use of machine learning tools is one of the goals of Google’s PAIR (People + AI Research) initiative, which was introduced in early July. Last w Read more…

By John Russell

Nvidia Responds to Google TPU Benchmarking

April 10, 2017

Nvidia highlights strengths of its newest GPU silicon in response to Google's report on the performance and energy advantages of its custom tensor processor. Read more…

By Tiffany Trader

GlobalFoundries Puts Wind in AMD’s Sails with 12nm FinFET

September 24, 2017

From its annual tech conference last week (Sept. 20), where GlobalFoundries welcomed more than 600 semiconductor professionals (reaching the Santa Clara venue Read more…

By Tiffany Trader

Leading Solution Providers

Graphcore Readies Launch of 16nm Colossus-IPU Chip

July 20, 2017

A second $30 million funding round for U.K. AI chip developer Graphcore sets up the company to go to market with its “intelligent processing unit” (IPU) in Read more…

By Tiffany Trader

Amazon Debuts New AMD-based GPU Instances for Graphics Acceleration

September 12, 2017

Last week Amazon Web Services (AWS) streaming service, AppStream 2.0, introduced a new GPU instance called Graphics Design intended to accelerate graphics. The Read more…

By John Russell

EU Funds 20 Million Euro ARM+FPGA Exascale Project

September 7, 2017

At the Barcelona Supercomputer Centre on Wednesday (Sept. 6), 16 partners gathered to launch the EuroEXA project, which invests €20 million over three-and-a-half years into exascale-focused research and development. Led by the Horizon 2020 program, EuroEXA picks up the banner of a triad of partner projects — ExaNeSt, EcoScale and ExaNoDe — building on their work... Read more…

By Tiffany Trader

Delays, Smoke, Records & Markets – A Candid Conversation with Cray CEO Peter Ungaro

October 5, 2017

Earlier this month, Tom Tabor, publisher of HPCwire and I had a very personal conversation with Cray CEO Peter Ungaro. Cray has been on something of a Cinderell Read more…

By Tiffany Trader & Tom Tabor

Cray Moves to Acquire the Seagate ClusterStor Line

July 28, 2017

This week Cray announced that it is picking up Seagate's ClusterStor HPC storage array business for an undisclosed sum. "In short we're effectively transitioning the bulk of the ClusterStor product line to Cray," said CEO Peter Ungaro. Read more…

By Tiffany Trader

Intel Launches Software Tools to Ease FPGA Programming

September 5, 2017

Field Programmable Gate Arrays (FPGAs) have a reputation for being difficult to program, requiring expertise in specialty languages, like Verilog or VHDL. Easin Read more…

By Tiffany Trader

IBM Advances Web-based Quantum Programming

September 5, 2017

IBM Research is pairing its Jupyter-based Data Science Experience notebook environment with its cloud-based quantum computer, IBM Q, in hopes of encouraging a new class of entrepreneurial user to solve intractable problems that even exceed the capabilities of the best AI systems. Read more…

By Alex Woodie

HPC Chips – A Veritable Smorgasbord?

October 10, 2017

For the first time since AMD's ill-fated launch of Bulldozer the answer to the question, 'Which CPU will be in my next HPC system?' doesn't have to be 'Whichever variety of Intel Xeon E5 they are selling when we procure'. Read more…

By Dairsie Latimer

  • arrow
  • Click Here for More Headlines
  • arrow
Share This