Today, data-intensive problems are solved using HPC resources, which are either large-scale Shared Memory Processing (SMP) systems, referred to as mainframes, or clusters comprising a large pool of inter-connected commodity servers. Mainframes constitute high performing and easy programmable solutions, but are only affordable to a fraction of the market. Cluster solutions have significantly lower costs, but also increased processing time and increased programming complexity. The increased complexity requires specialist developers and significantly limits the market opportunities.
The motivation for this project is the market demand for a currently non-existing solution that offers high performance and ease of use at a low cost. The project will address the challenges and disadvantages associated with cluster computing. By addressing the limitations of cluster systems whilst keeping the computational cost at the same level, we aim to provide HPC with mainframe capabilities at cluster prices.
The result from this project will constitute a worldwide new innovation with a huge market potential. It will be attractive to all industries handling Big Data problems or in the need for high performance computing capacity. Examples include life sciences such as genome projects, energy industry, medicine, space, and financial trading. By drastically reducing the cost per computation we expect to also reach new industries, which previously could not access HPC capacity. Thus, our ground-breaking global innovation represents a game-changing tool for big data problem solving across existing and new industry sectors.
Research Council of Norway (Business Sector Program)
- Simula Research Laboratory
- University of Oslo