Open AccessAdaptive Conflict Detection Algorithm Based on Rochester Software Transactional Memory
Author(s) -
Ying Li,
Junting Liu,
Fuxiang Gao
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1746/1/012050
Subject(s) - computer science , transactional leadership , software transactional memory , algorithm , transactional memory , string (physics) , value (mathematics) , signature (topology) , software , false positive rate , database transaction , artificial intelligence , mathematics , psychology , machine learning , operating system , social psychology , programming language , geometry , mathematical physics
When the transactional memory system detects conflicts, the more read-write addresses are, the higher the false positive rate of this algorithm is. This paper studies the problem and proposes a new signature based optimization algorithm, adaptive increase signature algorithm (AISA). The algorithm can calculate the saturation value of the number of read-write addresses, and dynamically adjust the size of bit string through the calculated saturation value, thus greatly reducing the false positive rate. The experimental results show that AISA can control the false positive rate at a low level on the basis of considering the space cost, and its effect is almost the same as that of 6 times of space, which will improve the performance of transactional memory system greatly.