SCIENCE

Samplify Systems announces the availability of Prism FP, the first compression technology to provide lossless and near lossless compression of floating point data types.  

"Distribution of very large data sets across hundreds, thousands or tens of thousands of computing nodes poses a challenge in many high-performance computing applications," says Al Wegener, Samplify Founder and CTO.  "With our Prism FP technology, users of high-performance computing applications can spend less time distributing their data sets, and more time processing them."

Users of high-performance computing applications like computational fluid dynamics, weather forecasting, automobile crash simulations, financial analysis, military, and genomics will benefit from Prism FP.  Indeed, any computing problem which involves solving partial differential equations using finite element analysis is a high-performance computing application, and a candidate for Prism FP.  High-performance computing, technical computing and supercomputing applications are all characterized by the use of a great number of computing nodes operating on data sets which exceed the local memory on any given node.  Distribution of these large 32-bit and 64-bit floating point data sets, among computing nodes and between computing nodes and storage arrays, drags down the performance of HPC applications.  Samplify's Prism FP compression technology reduces the time required for distribution of data sets and intermediate results, thereby reducing the execution time of HPC applications.

Prism FP will be available from Samplify during the second quarter of 2011 for Intel-based Linux environments.  Support for CUDA, OpenCL, MPI, and OpenMP environments will be provided later in 2011. Contact Samplify for development and per-node run-time licensing terms.