Projects details

  • LESS - LESS (is more): Low-level External Superoptimizer for Smart contracts
  • Jan 2021 - Apr 2022
  • Time and space are costly resources in smart contracts based blockchain. The compilers in use in smart contract blockchain are far from state of the art in regular compilation techniques. The purpose of the project is to create an optimizer for Michelson based on combinatorial optimization techniques, which are more suited to achieve significant code improvements. The challenges tackled by this TRA are general to programmable smart contracts based blockchain, and to low-level machine code optimization. Tezos and Michelson are the concrete testbed which we will base our findings on. 1. Identify optimization criteria/metrics 2. Identify methods and techniques to optimize Michelson code 3. Devise a generic framework for identity code optimization 4. Ensure scalability (code size and available computational resources) 5. Exploration of Metaheuristics and combinations thereof w.r.t. appropriateness for optimizing Michelson and similar low-level machine code The tools we will develop will be built on an existing framework for combinatorial optimization.
  • PN
  • NOVA LINCS / NOVA University of Lisbon
  • Fundação da Faculdade de Ciências e Tecnologia, UNL,NOVA LINCS / NOVA University of Lisbon,Universidade de Évora,Universidade da Beira Interior
  • 26,761.46
  • 26,761.46
  • 1 Jan 2021
  • 30 Apr 2022
  • Henrique João Domingos [Coordinator], Simão Melo de Sousa [Coordinator], Salvador Abreu [Coordinator], Henrique João Domingos [Researcher], Simão Melo de Sousa [Researcher], Salvador Abreu [Researcher]
  • NOVA LINCS / NOVA University of Lisbon
  • Fundação da Faculdade de Ciências e Tecnologia, UNL
  • NOVA LINCS / NOVA University of Lisbon
  • Universidade de Évora
  • Universidade da Beira Interior
  • The outcome of this TRA and its ensuing developments will improve on the state of the art in code superoptimization. The resulting tool will be based on a well-supported community framework for Constraint modeling and optimization, resorting (but not limited to) hybrid metaheuristics. Having a unique concrete low-level target (Michelson) will ease the realization of the goals for this project but the development will be designed so as to facilitate the application to different machine code ISAs and cost models.