Understanding object-level memory access patterns across the spectrum

Xu Ji, Chao Wang, Nosayba El-Sayed, Xiaosong Ma, Youngjae Kim, Sudharshan S. Vazhkudai, Wei Xue, Daniel Sanchez

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

Memory accesses limit the performance and scalability of countless applications. Many design and optimization efforts will benefit from an in-depth understanding of memory access behavior, which is not offered by extant access tracing and profiling methods. In this paper, we adopt a holistic memory access profiling approach to enable a better understanding of program-system memory interactions. We have developed a two-pass tool adopting fast online and slow offline profiling, with which we have profiled, at the variable/object level, a collection of 38 representative applications spanning major domains (HPC, personal computing, data analytics, AI, graph processing, and datacenter workloads), at varying problem sizes. We have performed detailed result analysis and code examination. Our findings provide new insights into application memory behavior, including insights on per-object access patterns, adoption of data structures, and memory-access changes at different problem sizes. We find that scientific computation applications exhibit distinct behaviors compared to datacenter workloads, motivating separate memory system design/optimizations.

Original languageEnglish
Title of host publicationProceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017
PublisherAssociation for Computing Machinery, Inc
ISBN (Electronic)9781450351140
DOIs
Publication statusPublished - 12 Nov 2017
Externally publishedYes
EventInternational Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017 - Denver, United States
Duration: 12 Nov 201717 Nov 2017

Other

OtherInternational Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017
CountryUnited States
CityDenver
Period12/11/1717/11/17

Fingerprint

Data storage equipment
Personal computing
Data structures
Scalability
Computer systems
Systems analysis
Processing

Keywords

  • Data types and structures
  • Memory profiling
  • Object access patterns
  • Tracing
  • Workload characterization

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Software

Cite this

Ji, X., Wang, C., El-Sayed, N., Ma, X., Kim, Y., Vazhkudai, S. S., ... Sanchez, D. (2017). Understanding object-level memory access patterns across the spectrum. In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017 [25] Association for Computing Machinery, Inc. https://doi.org/10.1145/3126908.3126917

Understanding object-level memory access patterns across the spectrum. / Ji, Xu; Wang, Chao; El-Sayed, Nosayba; Ma, Xiaosong; Kim, Youngjae; Vazhkudai, Sudharshan S.; Xue, Wei; Sanchez, Daniel.

Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017. Association for Computing Machinery, Inc, 2017. 25.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Ji, X, Wang, C, El-Sayed, N, Ma, X, Kim, Y, Vazhkudai, SS, Xue, W & Sanchez, D 2017, Understanding object-level memory access patterns across the spectrum. in Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017., 25, Association for Computing Machinery, Inc, International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017, Denver, United States, 12/11/17. https://doi.org/10.1145/3126908.3126917
Ji X, Wang C, El-Sayed N, Ma X, Kim Y, Vazhkudai SS et al. Understanding object-level memory access patterns across the spectrum. In Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017. Association for Computing Machinery, Inc. 2017. 25 https://doi.org/10.1145/3126908.3126917
Ji, Xu ; Wang, Chao ; El-Sayed, Nosayba ; Ma, Xiaosong ; Kim, Youngjae ; Vazhkudai, Sudharshan S. ; Xue, Wei ; Sanchez, Daniel. / Understanding object-level memory access patterns across the spectrum. Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017. Association for Computing Machinery, Inc, 2017.
@inproceedings{49282220709244c6ba2ff50c053b66cc,
title = "Understanding object-level memory access patterns across the spectrum",
abstract = "Memory accesses limit the performance and scalability of countless applications. Many design and optimization efforts will benefit from an in-depth understanding of memory access behavior, which is not offered by extant access tracing and profiling methods. In this paper, we adopt a holistic memory access profiling approach to enable a better understanding of program-system memory interactions. We have developed a two-pass tool adopting fast online and slow offline profiling, with which we have profiled, at the variable/object level, a collection of 38 representative applications spanning major domains (HPC, personal computing, data analytics, AI, graph processing, and datacenter workloads), at varying problem sizes. We have performed detailed result analysis and code examination. Our findings provide new insights into application memory behavior, including insights on per-object access patterns, adoption of data structures, and memory-access changes at different problem sizes. We find that scientific computation applications exhibit distinct behaviors compared to datacenter workloads, motivating separate memory system design/optimizations.",
keywords = "Data types and structures, Memory profiling, Object access patterns, Tracing, Workload characterization",
author = "Xu Ji and Chao Wang and Nosayba El-Sayed and Xiaosong Ma and Youngjae Kim and Vazhkudai, {Sudharshan S.} and Wei Xue and Daniel Sanchez",
year = "2017",
month = "11",
day = "12",
doi = "10.1145/3126908.3126917",
language = "English",
booktitle = "Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017",
publisher = "Association for Computing Machinery, Inc",

}

TY - GEN

T1 - Understanding object-level memory access patterns across the spectrum

AU - Ji, Xu

AU - Wang, Chao

AU - El-Sayed, Nosayba

AU - Ma, Xiaosong

AU - Kim, Youngjae

AU - Vazhkudai, Sudharshan S.

AU - Xue, Wei

AU - Sanchez, Daniel

PY - 2017/11/12

Y1 - 2017/11/12

N2 - Memory accesses limit the performance and scalability of countless applications. Many design and optimization efforts will benefit from an in-depth understanding of memory access behavior, which is not offered by extant access tracing and profiling methods. In this paper, we adopt a holistic memory access profiling approach to enable a better understanding of program-system memory interactions. We have developed a two-pass tool adopting fast online and slow offline profiling, with which we have profiled, at the variable/object level, a collection of 38 representative applications spanning major domains (HPC, personal computing, data analytics, AI, graph processing, and datacenter workloads), at varying problem sizes. We have performed detailed result analysis and code examination. Our findings provide new insights into application memory behavior, including insights on per-object access patterns, adoption of data structures, and memory-access changes at different problem sizes. We find that scientific computation applications exhibit distinct behaviors compared to datacenter workloads, motivating separate memory system design/optimizations.

AB - Memory accesses limit the performance and scalability of countless applications. Many design and optimization efforts will benefit from an in-depth understanding of memory access behavior, which is not offered by extant access tracing and profiling methods. In this paper, we adopt a holistic memory access profiling approach to enable a better understanding of program-system memory interactions. We have developed a two-pass tool adopting fast online and slow offline profiling, with which we have profiled, at the variable/object level, a collection of 38 representative applications spanning major domains (HPC, personal computing, data analytics, AI, graph processing, and datacenter workloads), at varying problem sizes. We have performed detailed result analysis and code examination. Our findings provide new insights into application memory behavior, including insights on per-object access patterns, adoption of data structures, and memory-access changes at different problem sizes. We find that scientific computation applications exhibit distinct behaviors compared to datacenter workloads, motivating separate memory system design/optimizations.

KW - Data types and structures

KW - Memory profiling

KW - Object access patterns

KW - Tracing

KW - Workload characterization

UR - http://www.scopus.com/inward/record.url?scp=85040185017&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85040185017&partnerID=8YFLogxK

U2 - 10.1145/3126908.3126917

DO - 10.1145/3126908.3126917

M3 - Conference contribution

BT - Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2017

PB - Association for Computing Machinery, Inc

ER -