Call for papers.

Synopsis High-performance interconnects have had a major role in high-performance scientific computing for several decades. They are also gaining traction in cloud computing and massive traditional data centers. The intersection of high-performance interconnects between these two areas spans networks of all scales, from small on-chip interconnects to whole system interconnects with hundreds of thousands of links. Innovation of new network architectures both in terms of software and hardware systems is a key component in next generation systems. As the number of cores in a single node increases, the bandwidth and message rates required from a single network interface controller (NIC) has continued to rise rapidly. Improvements to NIC architectures are required in order to keep up with increasing core counts on CPUs. In addition, improvements in network topologies have driven down costs by reducing the number of links and switches required to provide very high performance. These changes have also introduced interesting challenges as innovation in network topologies requires adopting adaptive routing and advanced congestion avoidance mechanisms. This special issue concentrates on the latest cutting edge developments in interconnects from both the cloud/data center and scientific computing communities.