Scalable FSM parallelization via path fusion and higher-order speculation

Qiu, Junqiao; Sun, Xiaofan; Sabet, Amir Hossein Nodehi; Zhao, Zhijia

doi:10.1145/3445814.3446705

Cited by 9 publications

(2 citation statements)

References 87 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Qiu Jun-qiao et al [53] have accomplished the parallelization of an extensible Finite States Machine using path fusion and high-order prediction. The concepts of path fusion and higher-order prediction are based on state enumeration and vector fusion, and speculative execution, respectively.…”

Section: Reducing the Number Of Local Searches In Iterative Algorithm...mentioning

confidence: 99%

“…Fig. 12 (a) Path Fusion (b)High-Order Prediction [53] Stephanus Daniel Handoko et al [52] achieved a maximum speedup of approximately 20.49x with an exhaustiveness of 8, and an average speedup of about 2.3x. In 2017, Yuguang Mu et al [54] introduced QuickVina-W, which enables researchers to efficiently and accurately screen large ligand libraries without prior definition of target pockets.…”

Section: Reducing the Number Of Local Searches In Iterative Algorithm...mentioning

confidence: 99%

See 1 more Smart Citation

A Review of the Parallelization Strategies for Iterative Algorithms

Zhou,

Ling,

Tang

et al. 2023

Preprint

View full text Add to dashboard Cite

Iteration-based algorithms have been widely used and achieved excellent results in many fields. However, in the big data era, data that needs to be processed is enormous in terms of both depth (the dimensionality of data) and breadth (the volume of data). Due to the slowdown of Moore's Law, the computing power of single-core CPUs is becoming saturated. The increase in the computational complexity and the bottleneck of the single-core processor’s speed exacerbate the time-consuming problem of iterative algorithms. With the rise of multi-core computers and distributed computing systems, parallelizing and deploying iterative algorithms on such systems can make full use of computing resources and accelerate iterative computation, providing a new idea for solving the aforementioned problems. However, due to the logical dependency between two consecutive iterations in an iterative algorithm, it is difficult to directly implement the concurrent computation of such algorithms. To this end, many studies have been conducted on the parallelization of iterative algorithms in both academia and industry. This paper aims to conduct an in-depth research and analysis of these parallelization strategies. Firstly, the abstract description and classification of iterative algorithms are given. Then four concurrency strategies for iterative algorithms are summarized, including logical units that can be intrinsically concurrently computed, multi-initial state parallel search strategy, data parallelism, and task parallelism. Finally, the paper detailed the convergence of parallel iterative algorithms, focusing on building the mathematical model of asynchronous iterative algorithms, and summarizing the convergence conditions of asynchronous iterative algorithms.

show abstract

Section: Reducing the Number Of Local Searches In Iterative Algorithm...mentioning

confidence: 99%

Section: Reducing the Number Of Local Searches In Iterative Algorithm...mentioning

confidence: 99%

A Review of the Parallelization Strategies for Iterative Algorithms

Zhou,

Ling,

Tang

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

One Automaton to Rule Them All: Beyond Multiple Regular Expressions Execution

Cicolini,

Carloni,

Santambrogio

et al. 2024

2024 IEEE/ACM International Symposium on Code Generation and Optimization (CGO)

View full text Add to dashboard Cite

Regular Expressions (REs) matching is crucial to identify strings exhibiting certain morphological properties in a data stream, resulting paramount in contexts such as deep packet inspection in computer security and genome analysis in bioinformatics. Yet, due to their intrinsic data-dependence characteristics, REs represent a complex computational kernel, and numerous solutions investigate pattern-matching efficiency in different directions. However, most of them lack a comprehensive ruleset optimization approach to truly push the pattern matching performance when considering multiple REs together. Thus, exploiting REs morphological similarities within the same dataset allows memory reduction when storing the patterns and drastically improves the dataset-matching throughput.Based on this observation, we propose the Multi-RE Finite State Automata (MFSA) that extends the Finite State Automata (FSA) model to improve REs parallelization by leveraging similarities within a specific application ruleset. We design a multi-level compilation framework to manage REs merging and optimization to produce MFSA(s). Furthermore, we extend iNFAnt algorithm for MFSAs execution with the novel iMFAnt engine. Our evaluation investigates the MFSA size-reduction impact and the execution throughput compared with the one of multiple FSA in both single-and multi-threaded configurations. This approach shows an average 71.95% compression in terms of states, introducing limited compilation time overhead. Besides, best iMFAnt achieves a geomean 5.99× throughput improvement and 4.05× speedup against single and multiple parallel FSAs.

show abstract

A GPU-accelerated Data Transformation Framework Rooted in Pushdown Transducers

Nguyen

Becchi

2022

2022 IEEE 29th International Conference on High Performance Computing, Data, and Analytics (HiPC)

View full text Add to dashboard Cite

Scalable FSM parallelization via path fusion and higher-order speculation

Cited by 9 publications

References 87 publications

A Review of the Parallelization Strategies for Iterative Algorithms

A Review of the Parallelization Strategies for Iterative Algorithms

One Automaton to Rule Them All: Beyond Multiple Regular Expressions Execution

A GPU-accelerated Data Transformation Framework Rooted in Pushdown Transducers

Contact Info

Product

Resources

About