Primeur magazine

Edition: flash - Issue: 2015-12-13

Hardware

ACMQUEUE published an article by researchers of INP in Grenoble, Fraser University and other, describing "Optimizing NUMA systems applications with Carrefour". They conclude that NUMA architecture is for scaling the processor count of today's server-class systems. In the near future, they expect systems to have even more NUMA nodes and more complicated NUMA topologies. Contrary to previous NUMA studies, the authors' experiments found that congestion causes the most serious NUMA problems. Congestion happens when the rate of requests to memory controllers or the rate of traffic over interconnects is too high, which causes excessive delays for memory accesses. It can be alleviated by balancing the traffic among multiple memory controllers and interconnect links. Read further...

Focus on Europe

Following the release of the ETP4HPC Strategic Research Agenda on November 24th 2015, the ETP4HPC organisation would like to receive feedback on this document from the HPC community. Hence, they released a public call for comments. During this process, any organisation or individual can express their view on this document. They will use these collected comments to prepare the next full version of our SRA, which is due to be published in July 2017. However, the current document will not be amended. Read further...