Co-Designing MPI Library and Applications for InfiniBand Clusters

Provided by: Ohio State University
Topic: Big Data
Format: PDF
The field of High Performance Computing (HPC) is forging ahead with complex and high performing system architectures. It is predicted that by the turn of the decade, the authors would have surpassed Exaflop level of computing power. In order to provide balanced system performance, it is necessary to design processor, memory hierarchies and network architecture and topology in a cohesive manner. \"Co-designing applications and communication libraries to leverage features of underlying communication network is imperative for achieving optimal performance on modern computing clusters\".

Find By Topic