Linearizability: a correctness condition for concurrent objects
ACM Transactions on Programming Languages and Systems (TOPLAS)
Approximate analysis of reader and writer access to a shared resource
SIGMETRICS '90 Proceedings of the 1990 ACM SIGMETRICS conference on Measurement and modeling of computer systems
Performance analysis of concurrent-read exclusive-write
SIGMETRICS '91 Proceedings of the 1991 ACM SIGMETRICS conference on Measurement and modeling of computer systems
Transactional memory: architectural support for lock-free data structures
ISCA '93 Proceedings of the 20th annual international symposium on computer architecture
Disjoint-access-parallel implementations of strong shared memory primitives
PODC '94 Proceedings of the thirteenth annual ACM symposium on Principles of distributed computing
Proceedings of the fourteenth annual ACM symposium on Principles of distributed computing
STOC '95 Proceedings of the twenty-seventh annual ACM symposium on Theory of computing
Transactional lock-free execution of lock-based programs
Proceedings of the 10th international conference on Architectural support for programming languages and operating systems
Multiple Reservations and the Oklahoma Update
IEEE Parallel & Distributed Technology: Systems & Technology
Universal Constructions for Large Objects
WDAG '95 Proceedings of the 9th International Workshop on Distributed Algorithms
Transparent Support for Wait-Free Transactions
WDAG '97 Proceedings of the 11th International Workshop on Distributed Algorithms
Software transactional memory for dynamic-sized data structures
Proceedings of the twenty-second annual symposium on Principles of distributed computing
Language support for lightweight transactions
OOPSLA '03 Proceedings of the 18th annual ACM SIGPLAN conference on Object-oriented programing, systems, languages, and applications
A flexible framework for implementing software transactional memory
Proceedings of the 21st annual ACM SIGPLAN conference on Object-oriented programming systems, languages, and applications
Software transactional memory for large scale clusters
Proceedings of the 13th ACM SIGPLAN Symposium on Principles and practice of parallel programming
Transactions are back---but are they the same?
ACM SIGACT News
A lazy snapshot algorithm with eager validation
DISC'06 Proceedings of the 20th international conference on Distributed Computing
A speculation-friendly binary search tree
Proceedings of the 17th ACM SIGPLAN symposium on Principles and Practice of Parallel Programming
| Hi-index | 0.00 |
One way that software transactional memory implementations attempt to reduce synchronization conflicts among transactions is by supporting different kinds of access modes. One such implementation, Dynamic Software Transactional Memory (DSTM), supports three kinds of memory access: WRITE access, which allows an object to be observed and modified, READ access, which allows an object to be observed but not modified, and RELEASE access, which allows an object to be observed for a limited duration.In this paper, we examine the relative performance of these modes for simple benchmarks on a small-scale multiprocessor. We find that on this platform and for these benchmarks, the READ and RELEASE access benchmarks do not substantially increase transaction throughput (and sometimes reduce it). We blame the extra bookkeeping inherent in these access modes.In response, we propose a new SNAP access mode. This mode provides almost the same behavior as RELEASE mode, but admits much more efficient implementations.