PhD Defense of Tariq AlTurkestani

KAUST PhD Student Tariq Alturkestani will be defending his thesis on I/O optimizations for scientific applications.

Maximizing I/O Bandwidth for Out-of-Core HPC Applications on Heterogeneous Large-Scale Systems

PhD Dissertation Defense by Tariq Alturkestani PhD Student, Computer Science

Thursday, July 9, 16:00 - 17:00,

Contact Person: Tariq Alturkestani

The execution rate of floating-point operations has typically increased by an order of magnitude every four years during the last 30 years of modern computing. This exponential growth in terms of computational power has benefited from the hardware technology scaling. However, memory and storage systems have not maintained the same rate of technology scaling. Scientific applications that are traditionally compute-bound have witnessed a paradigm shift toward an I/O-bound regime. This has been further exacerbated due to the limited on-node memory capacity, where I/O operations become a major bottleneck for scaling up critical simulations.

Tariq will present his software solution to tackle this daunting challenge of reducing expensive data motion that is key to maintaining cores busy and achieving the expected performance. Good luck Tariq!

Designs of Remote-Shared, Node-Local Burst Buffer, and Heterogeneous compute node on the Summit system

More details here.