Published December 2, 2010
HPC2 Helps Janya Data Miners Dig Deeper
Janya, a Western New York woman-owned, self-funded language technology company, provides semantic analysis tools and solutions to extract critical information from unstructured and semi-structured data, creating actionable intelligence. Organizations use the multilingual semantic processing capabilities to accelerate sharing and discovery of hidden knowledge within their existing information stores.
The tools are based on a state-of-the-art hybrid of machine learning and linguistic technologies. Janya staff will be using HPC2 resources at University at Buffalo's Center for Computational Research to discover how the tools can most effectively make use of massively parallel computing to accelerate and improve text and sentiment mining in key areas such as health and medical operations, health research, open source competitive intelligence for government and industry, and industrial research.
Janya staff will also be using the CCR supercomputer to analyze a significant amount of open source data, too large to examine with standard computing resources. The results of the research will be used to improve Janya's products for commercial and government use.
216 Intel Xeon E5-2650v2 nodes:
2.6 GHz per core,
8 cores per socket,
2 sockets per node,
8 FLOPS per cycle
333 GFLOPS/node peak
(72 TFLOPS total)
144 nodes with Mellanox FDR IB (56 Gbps)