A nanopass infrastructure for compiler education

Sarkar, Dipanwita; Waddell, Oscar; Dybvig, R. Kent

doi:10.1145/1016848.1016878

Cited by 12 publications

(11 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Tock is currently over 20,000 non-blank lines of Haskell code. Tock is a nanopass compiler (Sarkar et al 2004), meaning that its design consists of many (currently around 40) small passes that operate on the Abstract Syntax Tree (AST) of the program, each performing one simple operation, for example: making names unique, or checking that variables declared constant are not modified. A pass that makes names unique must traverse the entire AST, operating on all names.…”

Section: Motivationmentioning

confidence: 99%

Alloy

Brown

Sampson

2009

Proceedings of the 2nd ACM SIGPLAN Symposium on Haskell

View full text Add to dashboard Cite

Data-type generic programming can be used to traverse and manipulate specific parts of large heterogeneously-typed tree structures, without the need for tedious boilerplate. Generic programming is often approached from a theoretical perspective, where the emphasis lies on the power of the representation rather than on efficiency. We describe use cases for a generic system derived from our work on a nanopass compiler, where efficiency is a real concern, and detail a new generics approach (Alloy) that we have developed in Haskell to allow our compiler passes to traverse the abstract syntax tree quickly. We benchmark our approach against several other Haskell generics approaches and statistically analyse the results, finding that Alloy is fastest on heterogeneously-typed trees.

show abstract

Section: Motivationmentioning

confidence: 99%

Alloy

Brown

Sampson

2009

Proceedings of the 2nd ACM SIGPLAN Symposium on Haskell

View full text Add to dashboard Cite

show abstract

“…The Harlan compiler is written in Scheme as a nanopassstyle compiler [26]. Programs are read in as S-Expressions and compiled into C++ programs that use OpenCL.…”

Section: Compilationmentioning

confidence: 99%

Region-based memory management for GPU programming languages

Holk

Newton

Siek

et al. 2014

Proceedings of the 2014 ACM International Conference on Object Oriented Programming Systems Languages &Amp; Applications

View full text Add to dashboard Cite

Graphics Processing Units (GPUs) can effectively accelerate many applications, but their applicability has been largely limited to problems whose solutions can be expressed neatly in terms of linear algebra. Indeed, most GPU programming languages limit the user to simple data structures-typically only multidimensional rectangular arrays of scalar values. Many algorithms are more naturally expressed using higher level language features, such as algebraic data types (ADTs) and first class procedures, yet building these structures in a manner suitable for a GPU remains a challenge. We present a region-based memory management approach that enables rich data structures in Harlan, a language for data parallel computing. Regions enable rich data structures by providing a uniform representation for pointers on both the CPU and GPU and by providing a means of transferring entire data structures between CPU and GPU memory. We demonstrate Harlan's increased expressiveness on several example programs and show that Harlan performs well on more traditional data-parallel problems.

show abstract

“…An approach related to compilation by normalization is the nanopass compiler infrastructure of Sarkar et al [52], which advocates the design of compilers as a long pipeline of very small stages in order to enhance the understandability of the compiler in an educational context. Each stage transforms the program to an intermediate, more low-level form.…”

Section: Compilation By Normalizationmentioning

confidence: 99%

Code generation by model transformation: a case study in transformation modularity

et al. 2009

View full text Add to dashboard Cite

The realization of model-driven software development requires effective techniques for implementing code generators for domain-specific languages. This paper identifies techniques for improving separation of concerns in the implementation of generators. The core technique is code generation by model transformation, that is, the generation of a structured representation (model) of the target program instead of plain text. This approach enables the transformation of code after generation, which in turn enables the extension of the target language with features that allow better modularity in code generation rules. The technique can also be applied to 'internal code generation' for the translation of high-level extensions of a DSL to lower-level constructs within the same DSL using model-to-model transformations. This paper refines our earlier description of code generation by model transformation with an improved architecture for the composition of model-to-model normalization rules, solving the problem of combining type analysis and transformation. Instead of coarse-grained stages that alternate between normalization and type analysis, we have developed a new style of type analysis that can be integrated with normalizing transformations in a fine-grained manner. The normalization strategy has a simple extension interface and integrates non-local, context-sensitive transformation rules. We have applied the techniques in a realistic case study of domain-specific language engineering, i.e. the code generator for WebDSL, using Stratego, a high-level transformation language that integrates model-to-model, modelto-code, and code-to-code transformations.

show abstract

A nanopass infrastructure for compiler education

Cited by 12 publications

References 12 publications

Alloy

Alloy

Region-based memory management for GPU programming languages

Code generation by model transformation: a case study in transformation modularity

Contact Info

Product

Resources

About