Low-latency scalable locking algorithm for shared memory multiprocessors

Amitabh Dave, Nayeem Islam, Roy H. Campbell

Research output: Contribution to journalConference article

Abstract

In this paper we deal exclusively with spin locks for small- to medium-scale shared memory multiprocessors. The performance of a lock depends on the contention for the critical section being protected by the lock. The selection of the most appropriate lock for a critical section is difficult because the level of contention can vary dynamically. We verify that no current lock has the best performance for all critical sections. We present a new lock, called the shared array lock, which alleviates this problem. The shared array lock belongs to the class of locks based on queuing of processors waiting for the lock. Unlike existing queue-based locks which suffer from high latency in low contention situations, experimental results show that the shared array lock has low latency and good scalability.

Original languageEnglish (US)
Pages (from-to)10-17
Number of pages8
JournalIEEE Symposium on Parallel and Distributed Processing - Proceedings
StatePublished - Dec 1 1994
EventProceeedings of the 6th IEEE Symposium on Parallel and Distributed Processing - Dallas, TX, USA
Duration: Oct 26 1994Oct 29 1994

Fingerprint

Scalability
Data storage equipment

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Low-latency scalable locking algorithm for shared memory multiprocessors. / Dave, Amitabh; Islam, Nayeem; Campbell, Roy H.

In: IEEE Symposium on Parallel and Distributed Processing - Proceedings, 01.12.1994, p. 10-17.

Research output: Contribution to journalConference article

@article{9c480c8d743e4f46ac28b6e609c33079,
title = "Low-latency scalable locking algorithm for shared memory multiprocessors",
abstract = "In this paper we deal exclusively with spin locks for small- to medium-scale shared memory multiprocessors. The performance of a lock depends on the contention for the critical section being protected by the lock. The selection of the most appropriate lock for a critical section is difficult because the level of contention can vary dynamically. We verify that no current lock has the best performance for all critical sections. We present a new lock, called the shared array lock, which alleviates this problem. The shared array lock belongs to the class of locks based on queuing of processors waiting for the lock. Unlike existing queue-based locks which suffer from high latency in low contention situations, experimental results show that the shared array lock has low latency and good scalability.",
author = "Amitabh Dave and Nayeem Islam and Campbell, {Roy H.}",
year = "1994",
month = "12",
day = "1",
language = "English (US)",
pages = "10--17",
journal = "IEEE Symposium on Parallel and Distributed Processing - Proceedings",
issn = "1063-6374",

}

TY - JOUR

T1 - Low-latency scalable locking algorithm for shared memory multiprocessors

AU - Dave, Amitabh

AU - Islam, Nayeem

AU - Campbell, Roy H.

PY - 1994/12/1

Y1 - 1994/12/1

N2 - In this paper we deal exclusively with spin locks for small- to medium-scale shared memory multiprocessors. The performance of a lock depends on the contention for the critical section being protected by the lock. The selection of the most appropriate lock for a critical section is difficult because the level of contention can vary dynamically. We verify that no current lock has the best performance for all critical sections. We present a new lock, called the shared array lock, which alleviates this problem. The shared array lock belongs to the class of locks based on queuing of processors waiting for the lock. Unlike existing queue-based locks which suffer from high latency in low contention situations, experimental results show that the shared array lock has low latency and good scalability.

AB - In this paper we deal exclusively with spin locks for small- to medium-scale shared memory multiprocessors. The performance of a lock depends on the contention for the critical section being protected by the lock. The selection of the most appropriate lock for a critical section is difficult because the level of contention can vary dynamically. We verify that no current lock has the best performance for all critical sections. We present a new lock, called the shared array lock, which alleviates this problem. The shared array lock belongs to the class of locks based on queuing of processors waiting for the lock. Unlike existing queue-based locks which suffer from high latency in low contention situations, experimental results show that the shared array lock has low latency and good scalability.

UR - http://www.scopus.com/inward/record.url?scp=0028699170&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0028699170&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:0028699170

SP - 10

EP - 17

JO - IEEE Symposium on Parallel and Distributed Processing - Proceedings

JF - IEEE Symposium on Parallel and Distributed Processing - Proceedings

SN - 1063-6374

ER -