Demonstration venue · read-only. Every page can be browsed; the buttons that would change it are switched off. Create an account to run TaxoReview on your own data. A simulation from public publication records: the people listed were not invited and did not agree to review.
? Tutorial for this page

Yahong Sui

dblp:07/5802 · DBLP profile ↗
← Back
1ranked-venue papers
0as first author
0since 2021 · last 1999
—ORCID · none

Domains — the database's venue-derived domains; a paper can count in several

Systems, architecture and hardware · 1

Expertise — from the expertise taxonomy: the topics of the expert's papers under the CCF categories. A weight counts papers with recency: 1 for a paper about the topic, 0.3 when the topic is its context, halved every five years.

Computer architecture, parallel and distributed computing, and storage systems
1 paper
Distributed systems · 61% Memory systems · 30% Parallel and multicore computing · 9%

Topics — the 4 heaviest of 4, each with the papers that count most for it

TopicWeightPapersLastEvidence papers
Distributed systems
consistency protocols
0.011999
LOTEC: A Simple DSM Consistency Protocol for Nested Object Transactions · PODC 1999
Memory systems › shared memory
distributed shared memory
0.011999
LOTEC: A Simple DSM Consistency Protocol for Nested Object Transactions · PODC 1999
Distributed systems › consistency models
entry consistency
0.011999
LOTEC: A Simple DSM Consistency Protocol for Nested Object Transactions · PODC 1999
Parallel and multicore computing
parallel programming models
0.011999
LOTEC: A Simple DSM Consistency Protocol for Nested Object Transactions · PODC 1999
YearPublicationVenuePosition
1999 LOTEC: A Simple DSM Consistency Protocol for Nested Object Transactions
abstract
In this paper, we describe an efficient software-only Distributed Shared Memory (DSM) consistency protocol for an unconventional but important application domain -object transactional processing.Designers of transactional applications, while sensitive to performance issues, already accept significant overhead for the added functionality provided by transactions and, we speculate, would be willing to accept some small additional overhead for the benefit of "easy" distributed parallel execution.This is in contrast to the performance critical needs of the high-end scientific applications typically targetted by DSM system designers.While, after 10 years of performance enhancement, software DSMs are still considered inadequate by a large part of the parallel computing community, we believe they are now more than sufficiently mature to support distributed transactional environments.The key to making DSM attractive to the developers of transactional applications is ease of use.The much touted simplification offered by the use of shared memory instead of message passing is not enough.It must be possible to hide the needed synchronization operations from the programmer.To achieve this goal, we have developed a system for nested object transactions that supports the automatic insertion of synchronization primitives.These primitives are then used to drive the operation of an improved form of Bershad's Entry Consistency (EC) protocol to maintain memory consistency between the processors/nodes in a distributed computing system.The protocol described is also compatible with a number of performance enhancements for DSM systems that have been described in the literature.
Peter C. J. Graham, Yahong Sui
PODC2