The complexity of node architectures in supercomputers increases as we<br>cross petaflop milestones on the way towards Exascale. Increasing<br>levels of parallelism in multi– and many–core chips and emerging<br>heterogeneity of computational resources coupled with energy and memory<br>constraints force a reevaluation of our approaches towards operating<br>systems and runtime environments.<br>The ROSS workshop, to be held as a full–day meeting at the HPDC 2015<br>conference in Portland, Oregon, USA, focuses on principles and techniques<br>to design, implement, optimize, or operate runtime and operating systems<br>for supercomputers and massively parallel machines.<br>– OS and runtime system scalability on many–node and multi/many–core systems<br>– specialized OSs for Supercomputing<br>– distributed/hybrid/partitioned OSs and runtime systems for Supercomputing<br>– fault tolerance<br>– system noise analysis and prevention<br>– interaction between middleware, runtime system, and the OS<br>– modeling and performance analysis of runtime systems<br>– OS and runtime considerations for large–volume, high–performance I/O<br>– parallel job startup<br>– memory management and emerging memory technologies<br>– the role of OS and runtime system in minimizing power usage<br>– real–time considerations for Supercomputing<br>
Abbrevation
ROSS
City
PortlandOR
Country
United States
Deadline Paper
Start Date
End Date
Abstract