The growth of computing power of large-scale HPC clusters has lead to the need for large high-bandwidth I/O systems. However, directly writing the large and bursty datasets to centralized parallel file systems can raise significant I/O contention issues on storage servers. To meet this requirement, high speed, local storage layers – Burst Buffers are being deployed in next-generation supercomputers. In this paper, we introduce the design of a burst buffer library integrated at the MPI layer – Hyper Burst Buffer (HBB) designed for aggregated write operations.
Articles
Related Articles
June 6, 2022
pmAddr: a persistent memory centric computing architecture
Traditional Processor-centric computing architectures do not scale-out well, because servers do not share their local main...
Read More >
1 MIN READING
November 13, 2017
Low-PAPR OFDM-based waveform for fifth-generation cellular communications
We introduce a novel technique for improving PA efficiency and coverage of communication links, especially advantageous...
Read More >
1 MIN READING
May 9, 2024
A Balanced Distributed Cascode Power Amplifier With an Integrated Chebyshev Load Balancer for Full-Duplex Wireless Operation
This work proposes a fully integrated transmitter front end based on a balanced distributed cascode power...
Read More >
1 MIN READING