Time and space optimal implementations of atomic multi-writer register

Authors:
Amos Israeli;Amnon Shaham
Affiliations:
School for Computer Science and Mathematics, Netanya Academic College, Israel;Intel, Israel
Venue:
Information and Computation
Year:
2005

Citing 14
Cited 0

Axioms for memory access in asynchronous hardware systems

ACM Transactions on Programming Languages and Systems (TOPLAS) - The MIT Press scientific computation series
Axioms for concurrent objects

POPL '87 Proceedings of the 14th ACM SIGACT-SIGPLAN symposium on Principles of programming languages
Hierarchical correctness proofs for distributed algorithms

PODC '87 Proceedings of the sixth annual ACM Symposium on Principles of distributed computing
Impossibility and universality results for wait-free synchronization

PODC '88 Proceedings of the seventh annual ACM Symposium on Principles of distributed computing
Optimality of wait-free atomic multiwriter variables

Information Processing Letters
Atomic snapshots of shared memory

Journal of the ACM (JACM)
Some combinatorial aspects of time-stamp systems

European Journal of Combinatorics
On interprocess communication and the implementation of multi-writer atomic registers

Theoretical Computer Science
How to share concurrent wait-free variables

Journal of the ACM (JACM)
Bounded Concurrent Time-Stamping

SIAM Journal on Computing
Concurrent Reading While Writing

ACM Transactions on Programming Languages and Systems (TOPLAS)
Bounded time-stamps

Distributed Computing
Atomic shared register access by asynchronous hardware

SFCS '86 Proceedings of the 27th Annual Symposium on Foundations of Computer Science
Concurrent reading while writing II: The multi-writer case

SFCS '87 Proceedings of the 28th Annual Symposium on Foundations of Computer Science

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper addresses the wide gap in space complexity of atomic, multi-writer, multi-reader register implementations. While the space complexity of all previous implementations is linear, the lower bounds are logarithmic. We present three implementations which close this gap: The first implementation is sequential and its role is to present the idea and data structures used in the second and third implementations. The second and third implementations are both concurrent, the second uses multi-reader physical registers while the third uses single-reader physical registers. Both the second and third implementations are optimal with respect to the two most important complexity criteria: Their space complexity is logarithmic and their time complexity is linear.