Understanding the I/O Performance Gap Between Cori KNL and Haswell
eScholarship, University of California, 2021
Online
academicJournal
The Cori system at NERSC has two computepartitions with different CPU architectures: a 2,004 nodeHaswell partition and a 9,688 node KNL partition, whichranked as the 5th most powerful and fastest supercomputeron the November 2016 Top 500 list. The compute partitionsshare a common storage configuration, and understanding theIO performance gap between them is important, impactingnot only to NERSC/LBNL users and other national labs, butalso to the relevant hardware vendors and software developers.In this paper, we have analyzed performance of single coreand single node IO comprehensively on the Haswell and KNLpartitions, and have discovered the major bottlenecks, whichinclude CPU frequencies and memory copy performance. Wehave also extended our performance tests to multi-node IOand revealed the IO cost difference caused by network latency,buffer size, and communication cost. Overall, we have developeda strong understanding of the IO gap between Haswell and KNLnodes and the lessons learned from this exploration will guideus in designing optimal IO solutions in many-core era.
Titel: |
Understanding the I/O Performance Gap Between Cori KNL and Haswell
|
---|---|
Autor/in / Beteiligte Person: | Liu, J ; Koziol ; Tang, H ; Tessier, F ; Bhimji ; Cook, B ; Byna, S ; Austin, B ; Thakur, B ; lockwood ; Deslippe ; prabhat |
Link: | |
Veröffentlichung: | eScholarship, University of California, 2021 |
Medientyp: | academicJournal |
Schlagwort: |
|
Sonstiges: |
|