World's most powerful big data machines charted on Graph 500
IBM's BlueGene/Q dominates an emerging ranking of data processing supercomputers
So while a new Cray supercomputer took first place on theTop500, it was another machine, Lawrence Livermore National Laboratory's Sequoia, that proved to be the most adept at processing data intensive workloads on the Graph 500.
Such differences in ranking between the two scales highlight the changing ways in which the world's most powerful supercomputers are being used. An increasing number of high performance computing (HPC) machines are being put to work on data analysis, rather than the traditional duties of modeling and simulation.
"I look around the exhibit floor [of the Supercomputing 2012 conference], and I'm hard-pressed to find a booth that is not doing big data or analytics. Everyone has recognized that data is a new workload for HPC," said David Bader, a computational science professor at the Georgia Institute of Technology who helps oversee the Graph 500.
The Graph 500 was created to chart how well the world's largest computers handle such data intensive workloads. The latest edition of the list was released at the SC12 supercomputing conference, being held this week in Salt Lake City.
In a nutshell, the Graph 500 benchmark looks at "how fast [a system] can trace through random memory addresses," Bader said. With data intensive workloads, "the bottleneck in the machine is often your memory bandwidth rather than your peak floating point processing rate," he added.
The approach is markedly different than Top500. The well-known Top500 list relies on the Linpack benchmark, which was created in 1974. Linpack measures how effectively a supercomputer executes floating point operations, which are used for mathematically intensive computations such as weather modeling or other three dimensional simulations.
The Graph 500, in contrast, places greater emphasis on how well a computer can search through a large data set. "Big data has a lot of irregular and unstructured data sets, irregular accesses to memory, and much more reliance on memory bandwidth and memory transactions than on floating point performance," Bader said.
For the Graph 500 benchmark, the supercomputer is given a large set of data, called a graph. A graph is an interconnected set of data, such as a group of connected friends on a social network like Facebook. A graph consists of a set of vertices and edges, and in the social media context a vertex would be a person and the edge that person's connection to another person. Some vertices have many connections while many others have fewer. The computer is given a single vertex and is timed on how quickly it discovers all the other vertices in a graph, namely by following the edges.
- Securing Mobile App Data - Comparing Containers and App Wrappers Analysts agree that Mobile Device Management (MDM) is not enough when it comes to securing app data. Although it remains a critical component...
- Capabilities You Need in an IP Address Management Solution A mismanaged IP space can cripple an otherwise healthy network. Take a moment to understand what you need in an enterprise-ready IPAM solution.
- IPv6 Fundamentals IPv6 is needed to sustain the growth of the Internet. The transition from IPv4 will require planning and likely some degree of support...
- Optimize IT Performance & Availability: Four Steps to Establish Effective IT Management Baselines More than ever before, your company's ability to grow hinges on IT performance and availability. Download this how-to report on establishing IT baselines,...
- Accelerate your innovation with IBM Bluemix™ Join us for a webcast introducing the new IBM BluemixTM. IBM Bluemix (www.bluemix.net) is a developer oriented Platform as a Service (PaaS) environment...
- Maximizing Availability for the Modern Data Center Check out this information-packed resource center for help in maximizing the availability of your data center - from overcoming challenges to choosing the... All High Performance Computing White Papers | Webcasts