Ghent University Academic Bibliography

Advanced

The multi-program performance model: debunking current practice in multi-core simulation

Kenzo Van Craeynest UGent and Lieven Eeckhout UGent (2011) 2011 IEEE International symposium on Workload Characterization. p.26-37
abstract
Composing a representative multi-program multi-core workload is non-trivial. A multi-core processor can execute multiple independent programs concurrently, and hence, any program mix can form a potential multi-program workload. Given the very large number of possible multiprogram workloads and the limited speed of current simulation methods, it is impossible to evaluate all possible multi-program workloads. This paper presents the Multi-Program Performance Model (MPPM), a method for quickly estimating multiprogram multi-core performance based on single-core simulation runs. MPPM employs an iterative method to model the tight performance entanglement between co-executing programs on a multi-core processor with shared caches. Because MPPM involves analytical modeling, it is very fast, and it estimates multi-core performance for a very large number of multi-program workloads in a reasonable amount of time. In addition, it provides confidence bounds on its performance estimates. Using SPEC CPU2006 and up to 16 cores, we report an average performance prediction error of 2.3% and 2.9% for system throughput (STP) and average normalized turnaround time (ANTT), respectively, while being up to five orders of magnitude faster than detailed simulation. Subsequently, we demonstrate that randomly picking a limited number of multi-program workloads, as done in current pactice, can lead to incorrect design decisions in practical design and research studies, which is alleviated using MPPM. In addition, MPPM can be used to quickly identify multi-program workloads that stress multi-core performance through excessive conflict behavior in shared caches; these stress workloads can then be used for driving the design process further.
Please use this url to cite or link to this publication:
author
organization
year
type
conference
publication status
published
subject
keyword
computer architecture, performance modeling, workload characterization, multi-core processors
in
2011 IEEE International symposium on Workload Characterization
pages
26 - 37
publisher
IEEE
place of publication
New York, NY, USA
conference name
2011 IEEE International symposium on Workload Characterization (IISWC 2011)
conference location
Austi, TX, USA
conference start
2011-11-06
conference end
2011-11-08
Web of Science type
Proceedings Paper
Web of Science id
000299350700003
ISBN
9781457720642
DOI
10.1109/IISWC.2011.6114194
project
HPC-UGent: the central High Performance Computing infrastructure of Ghent University
language
English
UGent publication?
yes
classification
P1
copyright statement
I have transferred the copyright for this publication to the publisher
id
1977490
handle
http://hdl.handle.net/1854/LU-1977490
date created
2012-01-03 17:18:40
date last changed
2013-09-17 10:49:38
@inproceedings{1977490,
  abstract     = {Composing a representative multi-program multi-core workload is non-trivial. A multi-core processor can execute multiple independent programs concurrently, and hence, any program mix can form a potential multi-program workload. Given the very large number of possible multiprogram workloads and the limited speed of current simulation methods, it is impossible to evaluate all possible multi-program workloads. This paper presents the Multi-Program Performance Model (MPPM), a method for quickly estimating multiprogram multi-core performance based on single-core simulation runs. MPPM employs an iterative method to model the tight performance entanglement between co-executing programs on a multi-core processor with shared caches. Because MPPM involves analytical modeling, it is very fast, and it estimates multi-core performance for a very large number of multi-program workloads in a reasonable amount of time. In addition, it provides confidence bounds on its performance estimates. Using SPEC CPU2006 and up to 16 cores, we report an average performance prediction error of 2.3\% and 2.9\% for system throughput (STP) and average normalized turnaround time (ANTT), respectively, while being up to five orders of magnitude faster than detailed simulation. Subsequently, we demonstrate that randomly picking a limited number of multi-program workloads, as done in current pactice, can lead to incorrect design decisions in practical design and research studies, which is alleviated using MPPM. In addition, MPPM can be used to quickly identify multi-program workloads that stress multi-core performance through excessive conflict behavior in shared caches; these stress workloads can then be used for driving the design process further.},
  author       = {Van Craeynest, Kenzo and Eeckhout, Lieven},
  booktitle    = {2011 IEEE International symposium on Workload Characterization},
  isbn         = {9781457720642},
  keyword      = {computer architecture,performance modeling,workload characterization,multi-core processors},
  language     = {eng},
  location     = {Austi, TX, USA},
  pages        = {26--37},
  publisher    = {IEEE},
  title        = {The multi-program performance model: debunking current practice in multi-core simulation},
  url          = {http://dx.doi.org/10.1109/IISWC.2011.6114194},
  year         = {2011},
}

Chicago
Van Craeynest, Kenzo, and Lieven Eeckhout. 2011. “The Multi-program Performance Model: Debunking Current Practice in Multi-core Simulation.” In 2011 IEEE International Symposium on Workload Characterization, 26–37. New York, NY, USA: IEEE.
APA
Van Craeynest, Kenzo, & Eeckhout, L. (2011). The multi-program performance model: debunking current practice in multi-core simulation. 2011 IEEE International symposium on Workload Characterization (pp. 26–37). Presented at the 2011 IEEE International symposium on Workload Characterization (IISWC 2011), New York, NY, USA: IEEE.
Vancouver
1.
Van Craeynest K, Eeckhout L. The multi-program performance model: debunking current practice in multi-core simulation. 2011 IEEE International symposium on Workload Characterization. New York, NY, USA: IEEE; 2011. p. 26–37.
MLA
Van Craeynest, Kenzo, and Lieven Eeckhout. “The Multi-program Performance Model: Debunking Current Practice in Multi-core Simulation.” 2011 IEEE International Symposium on Workload Characterization. New York, NY, USA: IEEE, 2011. 26–37. Print.