Researchers in the Massachusetts Institute of Technology (MIT) Computer Science and Artificial Intelligence Laboratory (CSAIL) have designed a device that uses inexpensive flash storage to process huge graphs using a personal computer.
The team created a device consisting of a flash chip array and computation "accelerator" that helps flash achieve DRAM-like performance. The device runs on a "sort-reduce" algorithm that organizes all access requests for graph data into a sequential order that flash can access rapidly, obtaining kilobyte-sized chunks of thousands of requests simultaneously.
To test the new device, the researchers ran it against traditional high-performance systems to process large graphs, including the Web Data Commons Hyperlink Graph, with 3.5 billion nodes and 128 billion connecting lines. The traditional systems required a server with 128 gigabytes of DRAM, while the researchers achieved the same performance with a personal computer running two of their devices totaling 1 gigabyte of DRAM and 1 terabyte of flash.
By combining several devices, the team could process graphs of up to 4 billion nodes and 128 billion connecting lines, which no other system could handle on the 128-gigabyte server.
From MIT News
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA
No entries found