Near Memory Computing
What is NeMeCo?
NeMeCo is an interdisciplinary training and research project that targets the development of a power-efficient HPC system for Big data processing based on near-memory computing.
“Making Big data work for Europe” is an important ingredient of the European Commission’s Digital Agenda for Europe targeted at generating sustainable economic growth by moving towards a knowledge economy. Big data applications extract value out of huge amounts of data by searching for correlations that can be used to predict business trends, find the best medical treatment for diseases, perform financial risk management, determine the best locations to drill for oil and gas, fight crime, and for many other purposes. The key challenges for processing Big data are its large volume (amount of data), high velocity (speed of processing), and great variety (range of different data types) which all continue to increase. For example, while state-of-the-art “Big data” volumes are on the order of terabytes (1012) to petabytes (1015) this is expected to grow rapidly towards exabytes (1018) in the coming years.
Near-memory processing is one of the few real solutions to address the current scaling issues in HPC systems in order to realize exascale computers that are needed for near-future Big data workloads. However, near-memory computing is still in its infancy. Before it can be established as an essential component of HPC systems and be exploited for accelerating Big data workloads, multiple challenges have to be addressed. Besides the design of the near-memory computing device itself, this also includes its integration into the overall computer system architecture, how multiple near-memory computing devices can work together to scale to larger data volumes, and how such a hybrid system can be effectively programmed to maximize performance and minimize power consumption at the system level.
NeMeCo Community Impact
The current European computer industry is lagging behind in the world market. One of the major opportunities of the NeMeCo research training program is to bring Europe at the forefront in designing the next generations HPC systems that exploit near-memory computing concepts to realize a scaling path towards affordable Exascale computing in the next decade. Partners within the NeMeCo network are working across national borders towards implementing the European Higher Education Area, formalising the basis of mutual training recognition between the academic and non-academic sectors within the EU. Partners will recognise the research work carried out by ESRs, also when they visit other partner facilities somewhere in the EU.
The implementation of the NeMeCo interdisciplinary EID training programme will set international standards that serve as a model for new European graduate training programmes in NeMeCo related areas (from Big data applications to near-memory acceleration). This is necessary to compete with the highly structured PhD training programmes offered abroad, e.g. by American organisations, and to increase the attraction of Europe as a destination for obtaining the highest standards of training.
Gagandeep SinghAs ESR3 the focus of his research is on NeMeCo Architecture and Accelerator design.
Lorenzo CheliniAs ESR2 the focus of his research is on compilation techniques for NeMeCo architecture.
Stefano CordaaAs ESR1 the focus of his research is on applications and run-time optimizations for NeMeCo.
NeMeCo training objectives
Training object #1: enhance the attractiveness of careers in high-performance computer system design for Big data processing through
advanced training from international experts working on cutting edge technologies in state-of-the-art laboratory facilities.
Training object #3: develop researchers with a proven ability to transform abstract, fundamental ideas into commercial outcomes with excellent transferable skills that will be of life-long use across sectors.
Training object #4: create an active, life-long network of young researchers across sectors whose personal contacts, support and expertise will help Europe to deepen understanding of the design and programming issues of high performance computing systems for Big data workloads.
Training object #5: cascade expertise and spread good practice throughout Europe by personnel exchange and delivering researchers with potential to become academic or industrial leaders in the near future.