The COncurrency and LOcality Challenge
The current trend in the supercomputer industry is focused on providing more and more computational cores. The HPC industry is mainly compute-centric (tera/petaflops and now working for exaflops capacity), ignoring data placement and thread concurrency management. However, these aspects are proving increasingly relevant to maintaining progress in solving scientific and engineering problems with the help of data processing and simulations. As the volume of data handled by HPC applications is growing dramatically combined with an increasing computing hardware heterogeneity, data placement (resource allocation and access) and data management now appear to be a real barrier that prevents the applications from fully exploiting the power of modern HPC datacenters. Therefore, the objective of COLOC project is to design, implement and validate new approaches to break this barrier. The COLOC consortium also aims at providing software vendors and simulation service providers with methodologies and tools to develop and hone their applications to gain the most value from expensive and heterogeneous compute resources.