Richard Bunt - Academia.edu (original) (raw)

Papers by Richard Bunt

Measurement and analysis of locality phases in file referencing behaviour

Proceedings of the 1986 ACM SIGMETRICS joint international conference on Computer performance modelling, measurement and evaluation - SIGMETRICS '86/PERFORMANCE '86, 1986

A measure of program locality and its application

Proceedings of the 1984 ACM SIGMETRICS conference on Measurement and modeling of computer systems - SIGMETRICS '84, 1984

Page 1. A Measure of Program Locality and Its Application Richard B. Bnnt Jennifer M. Murphy Shik... more Page 1. A Measure of Program Locality and Its Application Richard B. Bnnt Jennifer M. Murphy Shikharesh Majnmdar Department of Computational Science University of Saskatchewan ABSTRACT Although the phenomenon ...

Mobile IP-based multicast as a service for mobile hosts

Second International Workshop on Services in Distributed and Networked Environments, 1995

... The authors would also like to thank F'rithvi-raj Narayanan who provided val... more ... The authors would also like to thank F'rithvi-raj Narayanan who provided valuable insight duriiiy the initial stages of the implementation work. References [AB931 A. Acharya and B. Ft. Badrinath. DP-livering multicast messages in network5 with mobile hostrs. In Proc. ...

A measure of program locahty and ~ts apphcations

Adaptive File Cache Management for Mobile Computing

Lecture Notes in Computer Science, 2002

By performing file system operations on cached copies of files mobile users reduce their need to ... more By performing file system operations on cached copies of files mobile users reduce their need to access files across the network. Effective cache management strategies will limit use of the network to a small number of read misses and periodic reintegration of changes (updates), and thus can shield the mobile user from changes in bandwidth that will have a significant impact on performance. This paper addresses adaptive approaches to file cache management in mobile environments. The results of a trace-driven simulation study demonstrate that adaptive adjustment of the cache block size can improve performance for mobile users at acceptable cost.

Model-Led Optimisation of a Geometric Multigrid Application

2013 IEEE 10th International Conference on High Performance Computing and Communications & 2013 IEEE International Conference on Embedded and Ubiquitous Computing, 2013

ABSTRACT This paper details the construction of an analytical performance model of HYDRA, a produ... more ABSTRACT This paper details the construction of an analytical performance model of HYDRA, a production nonlinear multigrid solver used by Rolls-Royce for computational fluid dynamics simulations. The model captures both the computational behaviour of HYDRA's key subroutines and the behaviour of its proprietary communication library, OPlus, with an absolute error consistently under 16% on up to 384 cores of an Intel X5650-based commodity cluster. We demonstrate how a performance model can be used to highlight performance bottlenecks and unexpected communication behaviours, thereby guiding code optimisation efforts. Informed by model predictions, we implement an optimisation in OPlus that decreases the communication and synchronisation time by up to 3.01 times and consequently improves total application performance by 1.41 times.

Write caching in distributed file systems

Proceedings of 15th International Conference on Distributed Computing Systems, 1995

... Kerhong C&en Richard B. Bunt Derek L. Eager Department of Computer Science Un... more ... Kerhong C&en Richard B. Bunt Derek L. Eager Department of Computer Science University of Saskatchewan Saskatoon, Canada S7N 5A9 ... Policies focusing on alnortizing the cost of a disk seek operation over multiple write-back opera-tions perform better at the server of a two ...

An Analysis of Space Priority Queueing in ATM Switches

Proceeding of 13th IEEE Annual International Phoenix Conference on Computers and Communications, 1994

Page 1. An Analysis of Space Priority Queueing in ATM Switches Karen L. Reid Department of Comput... more Page 1. An Analysis of Space Priority Queueing in ATM Switches Karen L. Reid Department of Computer Science University of Toronto Toronto, ON M5S 1A4 Abstract High-speed packet switched networks such as B-ISDN require ...

Using shim technology to monitor DCE runtime performance

Obtaining performance data for application or system software is typically difficult, especially ... more Obtaining performance data for application or system software is typically difficult, especially when the source code is not available. While popular techniques such as event trapping or monitoring external network traffic may provide some indication of performance, their various shortcomings restrict the domain of applicability. In this paper, we illustrate a new approach to obtaining performance data that appears to

Bradford-Zipf phenomena in computer systems

Our objective is to improve disk I/O performance in multi-disk systems supporting multiple concur... more Our objective is to improve disk I/O performance in multi-disk systems supporting multiple concur- rent users, such as file servers, database servers, and email servers. In such systems, many disk reads are absorbed by large in-memory buffers, and so disk writes comprise a large portion of the disk I/O traf- fic. LFS (Log-structured File System) has the po- tential to

Capacity bounds for cellular wireless mesh networks

A wireless transmitter can use a channel to transmit bits over distance per second (in general, e... more A wireless transmitter can use a channel to transmit bits over distance per second (in general, either more bits over shorter distance or fewer bits over longer distance). We call this the Channel Transport Capacity. This paper analyzes how the one-hop channel transport capacity limits multi-hop network capacity in cellular wireless mesh networks. The main results of this paper are

IEEE INFOCOM 2004, 2004

We present the results of an analysis of the usage of our new campus-wide wireless network. A wee... more We present the results of an analysis of the usage of our new campus-wide wireless network. A week-long traffic trace was collected in January 2003, recording address and protocol information for every packet sent and received on the wireless network. A centralised authentication log was used to match packets with wireless access points. The trace was analysed to answer questions about where, when, how much, and for what our wireless network is being used. Such information is important in evaluating design principles and planning for future network expansion.

Exploiting software interfaces for performance measurement

Proceedings of the first international workshop on Software and performance - WOSP '98, 1998

Software performance measurement can be a difficult and tedious procedure, and this difficulty ma... more Software performance measurement can be a difficult and tedious procedure, and this difficulty may explain the lack of interest shown in software performance optimisation in all but the most demanding areas, such as parallel computation and embedded systems. This paper describes the measurement shim, an approach to software perfor- mance which we have found to significantly reduce the effort required

An introduction to computer science

Taming the Unix Scheduler

Computer Measurement Group Conference, 1984

An experimental investigation of scheduling strategies for UNIX

Proceedings of the 1984 ACM SIGMETRICS conference on Measurement and modeling of computer systems - SIGMETRICS '84, 1984

The scheduler used in an operating system is an important factor in the performance of the system... more The scheduler used in an operating system is an important factor in the performance of the system under heavy load. This paper describes the scheduling philosophy employed in the UNIX operating system and outlines the standard ...

Disk cache replacement policies for network fileservers

[1993] Proceedings. The 13th International Conference on Distributed Computing Systems, 1993

Abstract Trace driven simulations were used to study the performance of several disk cache replac... more Abstract Trace driven simulations were used to study the performance of several disk cache replacement policies for network file servers. It is shown that locality based approaches, such as the common least recently used (LRU) policy, which are known to work well on ...

Characterising Program Behaviour with Phases and Transitions

Sigmetrics Performance Evaluation Review, 1988

Page 1. Characterising Program Behaviour With Phases and Transitions Jennifer M. Murphy Richard B... more Page 1. Characterising Program Behaviour With Phases and Transitions Jennifer M. Murphy Richard B. Bunt Department of Computational Science University of Saskatchewan ABSTRACT A detailed quantitative study of program behaviour is described. ...

Achieving Load Balance and Efiective Caching in Clustered Web Servers

This paper considers policies for distributing re- quests in clustered Web servers, wherein multi... more This paper considers policies for distributing re- quests in clustered Web servers, wherein multiple server machines are conflgured to function as a single high(er) performance Web server. We evaluate var- ious load distribution policies with respect to both their ability to achieve good load balance (the pri- mary goal) and also to their impact on the efiec- tiveness of per-machine

Measurement and analysis of locality phases in file referencing behaviour

Proceedings of the 1986 ACM SIGMETRICS joint international conference on Computer performance modelling, measurement and evaluation - SIGMETRICS '86/PERFORMANCE '86, 1986

A measure of program locality and its application

Proceedings of the 1984 ACM SIGMETRICS conference on Measurement and modeling of computer systems - SIGMETRICS '84, 1984

Page 1. A Measure of Program Locality and Its Application Richard B. Bnnt Jennifer M. Murphy Shik... more Page 1. A Measure of Program Locality and Its Application Richard B. Bnnt Jennifer M. Murphy Shikharesh Majnmdar Department of Computational Science University of Saskatchewan ABSTRACT Although the phenomenon ...

Mobile IP-based multicast as a service for mobile hosts

Second International Workshop on Services in Distributed and Networked Environments, 1995

... The authors would also like to thank F'rithvi-raj Narayanan who provided val... more ... The authors would also like to thank F'rithvi-raj Narayanan who provided valuable insight duriiiy the initial stages of the implementation work. References [AB931 A. Acharya and B. Ft. Badrinath. DP-livering multicast messages in network5 with mobile hostrs. In Proc. ...

A measure of program locahty and ~ts apphcations

Adaptive File Cache Management for Mobile Computing

Lecture Notes in Computer Science, 2002

By performing file system operations on cached copies of files mobile users reduce their need to ... more By performing file system operations on cached copies of files mobile users reduce their need to access files across the network. Effective cache management strategies will limit use of the network to a small number of read misses and periodic reintegration of changes (updates), and thus can shield the mobile user from changes in bandwidth that will have a significant impact on performance. This paper addresses adaptive approaches to file cache management in mobile environments. The results of a trace-driven simulation study demonstrate that adaptive adjustment of the cache block size can improve performance for mobile users at acceptable cost.

Model-Led Optimisation of a Geometric Multigrid Application

2013 IEEE 10th International Conference on High Performance Computing and Communications & 2013 IEEE International Conference on Embedded and Ubiquitous Computing, 2013

ABSTRACT This paper details the construction of an analytical performance model of HYDRA, a produ... more ABSTRACT This paper details the construction of an analytical performance model of HYDRA, a production nonlinear multigrid solver used by Rolls-Royce for computational fluid dynamics simulations. The model captures both the computational behaviour of HYDRA's key subroutines and the behaviour of its proprietary communication library, OPlus, with an absolute error consistently under 16% on up to 384 cores of an Intel X5650-based commodity cluster. We demonstrate how a performance model can be used to highlight performance bottlenecks and unexpected communication behaviours, thereby guiding code optimisation efforts. Informed by model predictions, we implement an optimisation in OPlus that decreases the communication and synchronisation time by up to 3.01 times and consequently improves total application performance by 1.41 times.

Write caching in distributed file systems

Proceedings of 15th International Conference on Distributed Computing Systems, 1995

... Kerhong C&en Richard B. Bunt Derek L. Eager Department of Computer Science Un... more ... Kerhong C&en Richard B. Bunt Derek L. Eager Department of Computer Science University of Saskatchewan Saskatoon, Canada S7N 5A9 ... Policies focusing on alnortizing the cost of a disk seek operation over multiple write-back opera-tions perform better at the server of a two ...

An Analysis of Space Priority Queueing in ATM Switches

Proceeding of 13th IEEE Annual International Phoenix Conference on Computers and Communications, 1994

Page 1. An Analysis of Space Priority Queueing in ATM Switches Karen L. Reid Department of Comput... more Page 1. An Analysis of Space Priority Queueing in ATM Switches Karen L. Reid Department of Computer Science University of Toronto Toronto, ON M5S 1A4 Abstract High-speed packet switched networks such as B-ISDN require ...

Using shim technology to monitor DCE runtime performance

Obtaining performance data for application or system software is typically difficult, especially ... more Obtaining performance data for application or system software is typically difficult, especially when the source code is not available. While popular techniques such as event trapping or monitoring external network traffic may provide some indication of performance, their various shortcomings restrict the domain of applicability. In this paper, we illustrate a new approach to obtaining performance data that appears to

Bradford-Zipf phenomena in computer systems

Our objective is to improve disk I/O performance in multi-disk systems supporting multiple concur... more Our objective is to improve disk I/O performance in multi-disk systems supporting multiple concur- rent users, such as file servers, database servers, and email servers. In such systems, many disk reads are absorbed by large in-memory buffers, and so disk writes comprise a large portion of the disk I/O traf- fic. LFS (Log-structured File System) has the po- tential to

Capacity bounds for cellular wireless mesh networks

A wireless transmitter can use a channel to transmit bits over distance per second (in general, e... more A wireless transmitter can use a channel to transmit bits over distance per second (in general, either more bits over shorter distance or fewer bits over longer distance). We call this the Channel Transport Capacity. This paper analyzes how the one-hop channel transport capacity limits multi-hop network capacity in cellular wireless mesh networks. The main results of this paper are

IEEE INFOCOM 2004, 2004

We present the results of an analysis of the usage of our new campus-wide wireless network. A wee... more We present the results of an analysis of the usage of our new campus-wide wireless network. A week-long traffic trace was collected in January 2003, recording address and protocol information for every packet sent and received on the wireless network. A centralised authentication log was used to match packets with wireless access points. The trace was analysed to answer questions about where, when, how much, and for what our wireless network is being used. Such information is important in evaluating design principles and planning for future network expansion.

Exploiting software interfaces for performance measurement

Proceedings of the first international workshop on Software and performance - WOSP '98, 1998

Software performance measurement can be a difficult and tedious procedure, and this difficulty ma... more Software performance measurement can be a difficult and tedious procedure, and this difficulty may explain the lack of interest shown in software performance optimisation in all but the most demanding areas, such as parallel computation and embedded systems. This paper describes the measurement shim, an approach to software perfor- mance which we have found to significantly reduce the effort required

An introduction to computer science

Taming the Unix Scheduler

Computer Measurement Group Conference, 1984

An experimental investigation of scheduling strategies for UNIX

Proceedings of the 1984 ACM SIGMETRICS conference on Measurement and modeling of computer systems - SIGMETRICS '84, 1984

The scheduler used in an operating system is an important factor in the performance of the system... more The scheduler used in an operating system is an important factor in the performance of the system under heavy load. This paper describes the scheduling philosophy employed in the UNIX operating system and outlines the standard ...

Disk cache replacement policies for network fileservers

[1993] Proceedings. The 13th International Conference on Distributed Computing Systems, 1993

Abstract Trace driven simulations were used to study the performance of several disk cache replac... more Abstract Trace driven simulations were used to study the performance of several disk cache replacement policies for network file servers. It is shown that locality based approaches, such as the common least recently used (LRU) policy, which are known to work well on ...

Characterising Program Behaviour with Phases and Transitions

Sigmetrics Performance Evaluation Review, 1988

Page 1. Characterising Program Behaviour With Phases and Transitions Jennifer M. Murphy Richard B... more Page 1. Characterising Program Behaviour With Phases and Transitions Jennifer M. Murphy Richard B. Bunt Department of Computational Science University of Saskatchewan ABSTRACT A detailed quantitative study of program behaviour is described. ...

Achieving Load Balance and Efiective Caching in Clustered Web Servers

This paper considers policies for distributing re- quests in clustered Web servers, wherein multi... more This paper considers policies for distributing re- quests in clustered Web servers, wherein multiple server machines are conflgured to function as a single high(er) performance Web server. We evaluate var- ious load distribution policies with respect to both their ability to achieve good load balance (the pri- mary goal) and also to their impact on the efiec- tiveness of per-machine