Sponsored By
In Cooperation With
Gold Sponsors
Bronze Sponsors
Logo design by Michal O. Zadok.
|
Xiaosong Ma - Mohamed bin Zayed University of Artificial IntelligenceSubject: I/O Coordination for Better Resource Sharing -- From HPC to AI Storage Abstract: Despite tremendous improvement in absolute capacity and speed, the storage subsystem remains one of the slower, less predictable, and less scalable components of large-scale applications. In this talk, through a personal journey of parallel and distributed storage systems, I hope to share observations and lessons from these past projects. In particular, across the many layers of the storage hierarchy studied, from the CPU caches to supercomputer and cloud storage clusters, we repeatedly encounter the same underlying challenge of efficient sharing of I/O resources. Without effective regulation of I/O flows, the storage system can easily be a performance bottleneck while remaining largely underutilized. The discussion of solutions continues into related new (and old) storage problems with today's parallel AI training and inference workloads.Bio: Professor Ma’s main research focus has been on efficient resource utilization across different system layers, from CPU/GPU caches to datacenter-scale distributed storage. Her research projects share the common theme of reducing or eliminating waste in systems software, to better support applications including graph, database, and LLM processing. Prior to joining Mohamed bin Zayed University of Artificial Intelligence, Professor Ma worked as a principal scientist at Qatar Computing Research Institute, and as an associate professor at North Carolina State University. She has produced 100 publications and served as associate editor for ACM ToS and Elsevier JPDC, as well as PC co-chair for conferences such as USENIX FAST 2024. In addition, she has served as PC member for numerous conferences, including OSDI, ASPLOS, FAST, EuroSys, Supercomputing, USENIX ATC, ICS, and HPDC. |