Generating an efficient compiler for a data parallel language from a denotational specification

In this paper we report on a long-term research effort to develop and use efficient language implementation generators in practice. The generator is applied to a number of different languages, some of which are used for projects in industry. The used formal specification style is Operational Semantics, primarily in the form called Natural Semantics, represented and supported by a meta-language and tool called the Relational Meta Language (RML), which can generate efficient implementations in C, on par with hand-implemented code.Generating implementations from formal specifications are assumed to give advantages such as: high level descriptions, higher degree of correctness, and consistency between specification and implementation. To what extent can this be realized in practice? Does it scale to large language implementations?To answer some of these questions we have developed specifications of a range of languages: imperative, functional, object-oriented (Java), and equation-based (Modelica). The size of specifications range from half a page to large specifications of 60 000 lines. It turns out to be possible to generate efficient compilers, also for large languages. However, the performance of the generator tool and the user support of the development environment become increasingly important for large specifications. To satisfy such user needs the speed of the generator was increased a factor of ten to reduce turn-around time, and an Eclipse plug-in including a debugger were developed. For very large specifications, the structuring and modularity of the specification itself also become essential for performance and maintainability.

show abstract

Automatic generation of provably correct parallelizing compilers

Gupta

Pontelli²,

Lara-Rodriguez³

et al.

Proceedings. 1998 International Conference on Parallel Processing (Cat. No.98EX205)

View full text Add to dashboard Cite

We show how parallelizing compilers can be automatically derived from denotational de nitions of programming languages. In our approach, the denotational de nition is expressed using de nite clause grammars (syntax speci cation) and Horn Logic or Constraint Logic (semantic speci cation). The conditions for executing two or more statements in parallel (e.g., GCD test, Banerjee test, or exact test) are included as part of the (parallel) denotational semantics of the language. Solutions of diophantine equations, needed for parallelizing DO loops, can be expressed in constraint logic as well, and are thus easily incorporated in our denotational framework. This parallel denotational speci cation of the language is executable, and thus automatically yields a parallel interpreter. This interpreter can be partially evaluated w.r.t. a given program to obtain parallel compiled code. In addition, the various syntactic and semantic restructuring transformations that have been proposed to expose more parallelism in sequential programs can also be expressed in our denotational framework. Partial evaluation of a program w.r.t. this interpreter will result in parallel code that is specialized to a given architecture. Automatic derivation of these parallelizing compilers becomes possible because of use of Horn Logic, instead of the traditional -calculus, for expressing denotational semantics. The major advantage of our approach is that the parallelizing compilers derived are provably correct, since they are automatically generated from language speci cations.

show abstract

Generating an efficient compiler for a data parallel language from a denotational specification

Cited by 3 publications

References 19 publications

Horn Logic Denotations and Their Applications

Horn Logic Denotations and Their Applications

Formal Semantics Based Translator Generation and Tool Development in Practice

Automatic generation of provably correct parallelizing compilers

Contact Info

Product

Resources

About