Rethinking supercompilation

Mitchell, Neil

doi:10.1145/1932681.1863588

Cited by 9 publications

(27 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the higher-order formulations of positive supercompilation given by Mitchell [13] and Bolingbroke [2], all expressions are memoised. The extra work required for the additional computationally expensive homeomorphic embedding checks is very time consuming, which has been borne out by experimental results.…”

Section: Discussionmentioning

confidence: 99%

“…Also, memoising additional expressions to ensure termination may result in a loss of program efficiency since new function calls may be introduced without being offset by corresponding function unfoldings. It should also be pointed out that the implementation of positive supercompilation in [13] will not terminate on programs such as the second one in Section 1. This is because the simplification rules that are applied to terms prior to transformation by positive supercompilation will not terminate for such programs which use contravariant (negative) data types.…”

Section: Discussionmentioning

confidence: 99%

“…Later it has been used for higher-order positive supercompilation-see [13,2,6,12,7,8,9,5]. In some of these works two expressions must be coupled at the outer-most level to avoid the split-operation; but we will need that operation anyway.…”

Section: Generalization: Whenmentioning

confidence: 99%

“…A specific variant of the transformation techinque was later popularized in the setting of a more traditional functional language in a form known as positive supercompilation [14,17]. More recently, it has been generalized to higher-order languages in several new lines of work [13,2,7,8,9,5].…”

Section: Introductionmentioning

confidence: 99%

“…To avoid non-termination, some formulations of positive supercompilation for a higher-order language memoise all expressions [13,2], or at least a substantial subset of them [7,8,9,5]. However, this may be a costly operation during transformation and may result in less optimized output programs (that may actually be less efficient than the input programs).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Local Driving in Higher-Order Positive Supercompilation via the Omega-theorem

Hamilton¹,

Sørensen²

EPiC Series in Computing

View full text Add to dashboard Cite

A program transformation technique should terminate, return efficient output programs and be efficient itself. These requirements are mutually conflicting, so a balance must be sought between definite termination and possible efficiency.For positive supercompilation [17], ensuring termination requires memoisation of expressions, and these are subsequently used to determine when to perform generalization and folding [16]. For a first-order language, every infinite sequence of transformation steps must include function unfolding, so it is sufficient to memoise only those expressions immediately prior to a function unfolding step.However, for a higher-order language, it is possible for an expression to have an infinite sequence of transformation steps which do not include function unfolding, so memoisation prior to a function unfolding step is not sufficient by itself to ensure termination. But memoising additional expressions is expensive during transformation and may lead to less efficient output programs due to auxiliary functions. This additional memoisation may happen explicitly during transformation or implicitly via a pre-processing transformation as outlined in previous work by the first author [5]. We introduce a new technique for local driving in higher-order positive supercompilation which obliviates the need for memoising other expressions than function unfolding steps, thereby improving efficiency of both the transformation and the generated programs. We exploit the fact, due to the second author in the setting of type-free λ-calculus [20] known as the Ω-theorem, that every expression with an infinite sequence of transformation steps not involving function unfolding must have the term Ω = (λx.x x) (λx.x x) embedded within it in a certain sense. The technique has proven useful on a host of examples.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Generalization: Whenmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Local Driving in Higher-Order Positive Supercompilation via the Omega-theorem

Hamilton¹,

Sørensen²

EPiC Series in Computing

View full text Add to dashboard Cite

show abstract

Supercompilation by evaluation

Jones

2010

Proceedings of the Third ACM Haskell Symposium on Haskell

View full text Add to dashboard Cite

Supercompilation is a technique due to Turchin [1] which allows for the construction of program optimisers that are both simple and extremely powerful. Supercompilation is capable of achieving transformations such as deforestation [2], function specialisation and constructor specialisation [3]. Inspired by Mitchell's promising results [4], we show how the call-by-need supercompilation algorithm can be recast to be based explicitly on an evaluator, and in the process extend it to deal with recursive let expressions.

show abstract

Taming code explosion in supercompilation

Jönsson

Nordlander

2011

Proceedings of the 20th ACM SIGPLAN Workshop on Partial Evaluation and Program Manipulation

View full text Add to dashboard Cite

Supercompilation algorithms can perform great optimizations but sometimes suffer from the problem of code explosion. This results in huge binaries which might hurt the performance on a modern processor. We present a supercompilation algorithm that is fast enough to speculatively supercompile expressions and discard the result if it turned out bad. This allows us to supercompile large parts of the imaginary and spectral parts of nofib in a matter of seconds while keeping the binary size increase below 5%.

show abstract

Rethinking supercompilation

Cited by 9 publications

References 26 publications

Local Driving in Higher-Order Positive Supercompilation via the Omega-theorem

Local Driving in Higher-Order Positive Supercompilation via the Omega-theorem

Supercompilation by evaluation

Taming code explosion in supercompilation

Contact Info

Product

Resources

About