SMT-COMP 2024: Final Call for Solvers

10 views
Skip to first unread message

Martin Bromberger

unread,
May 13, 2024, 10:24:48 AMMay 13
to smt-...@cs.nyu.edu, smt...@googlegroups.com, smt-an...@googlegroups.com

================================================



 18th International Satisfiability Modulo Theories Competition


                         (SMT-COMP'24)



               FINAL CALL FOR SOLVERS



=================================================



We now invite registration of solvers for SMT-COMP 2024.



Solvers are entered into the competition via a push request to the SMT-COMP GitHub repository at 


https://github.com/SMT-COMP/smt-comp.github.io/tree/master/submissions


Detailed instructions can be found in the README file there. Note that solvers must be uploaded to a publicly available space as a docker file that compiles the solver and the final version must be a Zenodo (https://zenodo.org/) submission for reproducibility.


The submission deadline for (first versions of) solvers is


    ***May 27, 2024. ***


After the above date, no new entrants will be accepted. However,

submitted solvers may be updated via a push request until


    ***June 8, 2024. ***


We plan to publish the results including all materials in an artifact.

By participating, you agree that we may copy your solver binaries to

this artifact.  Please include a license file of your choice with your

solver binaries.  Please contact us, if you want to include other files

with the artifact.


Note that a short system description of 1-2 pages (see Section 4 of the

competition rules at https://smt-comp.github.io/2024/rules.pdf) is part

of the solver submission and MANDATORY. Submission of the system

description is THIS YEAR due for the initial solver deadline on May 27,

2024. Participants are asked to provide a link to the system

description in the registration form. The system description can be

updated until June 8, 2024, however it must be reviewable by the

initial solver deadline. The links to the final solver submission must be 

provided via a push request to the organizers by the final deadline.


The single query, incremental, unsat-core, and model-validation tracks

will no longer run on StarExec (http://www.starexec.org) because 

StarExec is in the process of being decommissioned. Starting this year, 

we will run these tracks on the BenchExec cluster 

(https://github.com/sosy-lab/benchexec) owned by LMU's Software and 

Computational Systems Lab (SoSy-Lab https://www.sosy-lab.org/), who 

are kind enough to support our competition with their computing power.

To be more precise, the competition will be run on the 168 apollon nodes

of the SoSy-Lab BenchExec cluster (for more details see

https://vcloud.sosy-lab.org/cpachecker/webclient/master/info).

It is also possible to locally emulate and test the computing environment

on the competition machines using the following instructions:

https://gitlab.com/sosy-lab/benchmarking/competition-scripts/#computing-environment-on-competition-machines


You can use last year’s pre-/post-processors for testing. Note that for 

testing your solver in the incremental track you need to wrap the trace 

executor around your solver, see

https://github.com/smt-comp/trace-executor#wrapping-your-solver.

If you don’t wrap it yourself, we will wrap your solver in the competition.


Your solver can have multiple configurations for different tracks, logics, etc.  

These can be specified as part of the solver submission and changed 

until the deadline for the final solver.  The default configuration is used for 

all other tracks.


As in the last year, we organize a separate parallel and cloud solving

track, hosted by Amazon Web Services.  For more information on the

participation on these two tracks, see

https://smt-comp.github.io/2024/parallel-and-cloud-tracks.html.


Finally, we have decided to discontinue the proof exhibition track.

There are three reasons that led to this decision. First, we were

unable to find a way of turning the proof track into an actual competition

that would not lead to an unmanageable amount of overhead on the

competition organizers or the participants. Second, we have collected

vast amounts of data in the last two years and thoroughly evaluating it

takes time. Third, this year we would need to severely restrict the 

allowed proof size due to hardware limits of the new execution platform.


Please see the competition rules for further details.


Sincerely,


The organizing team

Martin Bromberger (chair), MPI for Informatics, Germany

François Bobot, CEA List, France

Martin Jonáš, Masaryk University, Czech Republic

Reply all
Reply to author
Forward
0 new messages