An Xdata Architecture for Federated Graph Models and Multi-tier Asymmetric Computing
Abstract
Scalable, data-parallel graph analytics on GPUs is a fundamentally hard problem that goes beyond the current state of the art. Scalable graph analytics are critical for a large range of application domains with a vital impact on both national security and the national economy. CPU graph algorithms are known to scale poorly due to non-locality and limited memory bandwidth. Our research shows that GPUs provide a high-performance, data-parallel, commodity hardware platform for graph analytics.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2014
- Accession Number
- ADA598040
Entities
People
- Bryan Thompson
- Michael Personick