Clause Sharing and Partitioning for Cloud-Based SMT Solving

Marescotti, Matteo; Hyvärinen, Antti E. J.; Sharygina, Natasha

doi:10.1007/978-3-319-46520-3_27

Cited by 11 publications

(9 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Instantiating parallelization trees on different domains requires domainspecific knowledge, in particular to properly define partitioning, portfolio and lemma sharing. In the following we provide an overview of the SMT and IC3 instantiations, and we refer the reader respectively to [17], and [16] for further details.…”

Section: Smt and Ic3 Instantiationsmentioning

confidence: 99%

“…In this section we report experimental results over both parallel SMT and IC3 using SMTS within a distributed environment. A more in-depth analysis on the techniques is available at [17] and [16], respectively. For all our experiments we use a modern Intel-based cluster of 24 nodes, each with 64GB of memory and 20 cores.…”

Section: Experimental Evaluationsmentioning

confidence: 99%

“…These are commonly used in combination with sharing of learned information. We have observed analytically for SMT [10] and in practical experiments for both SMT [17] and IC3 [16] that when used alone, both partitioning and algorithm portfolios have their own shortcomings and one often seems to perform well exactly when the other performs badly. The parallelization tree is a hybrid approach capable of exploiting the strengths of each parallel approach in such circumstances.…”

Section: Introductionmentioning

confidence: 97%

“…Related work. The parallelization tree formalism was initially proposed in [11], adapted to SMT in [9,17], and analysed further in [10]. The seminal paper on the model-checking algorithm IC3 [3] already provides experimental evidence on the efficiency of parallelization.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

SMTS: Distributed, Visualized Constraint Solving

Marescotti

Hyvärinen

Sharygina

EPiC Series in Computing

Self Cite

View full text Add to dashboard Cite

The inherent complexity of parallel computing makes development, resource monitor- ing, and debugging for parallel constraint-solving-based applications difficult. This paper presents SMTS, a framework for parallelizing sequential constraint solving algorithms and running them in distributed computing environments. The design (i) is based on a gen- eral parallelization technique that supports recursively combining algorithm portfolios and divide-and-conquer with the exchange of learned information, (ii) provides monitoring by visually inspecting the parallel execution steps, and (iii) supports interactive guidance of the algorithm through a web interface. We report positive experiences on instantiating the framework for one SMT solver and one IC3 solver, debugging parallel executions, and visualizing solving, structure, and learned clauses of SMT instances.

show abstract

Section: Smt and Ic3 Instantiationsmentioning

confidence: 99%

Section: Experimental Evaluationsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 97%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

SMTS: Distributed, Visualized Constraint Solving

Marescotti

Hyvärinen

Sharygina

EPiC Series in Computing

Self Cite

View full text Add to dashboard Cite

show abstract

“…An algorithm similar to Tree presented in this work was first used in parallel SAT solving approaches based on divide-and-conquer in [10], and later for SMT solving in [12,14]. However, the lookahead-based partitioning algorithm was never published for lack of space.…”

Section: Related Workmentioning

confidence: 99%

Lookahead-Based SMT Solving

Hyvärinen

Marescotti

Sadigova

et al.

EPiC Series in Computing

Self Cite

View full text Add to dashboard Cite

The lookahead approach for binary-tree-based search in constraint solving favors branching that provide the lowest upper bound for the remaining search space. The approach has recently been applied in instance partitioning in divide-and-conquer-based parallelization, but in general its connection to modern, clause-learning solvers is poorly understood. We show two ways of combining lookahead approach with a modern DPLL(T)-based SMT solver fully profiting from theory propagation, clause learning, and restarts. Our thoroughly tested prototype implementation is surprisingly efficient as an independent SMT solver on certain instances, in particular when applied to a non-convex theory, where the lookahead-based implementation solves 40% more unsatisfiable instances compared to the standard implementation.

show abstract