Shallow embedding of DSLs via online partial evaluation

Leißa, Roland; Boesche, Klaas; Hack, Sebastian; Membarth, Richard; Slusallek, Philipp

doi:10.1145/2814204.2814208

Cited by 13 publications

(7 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In fact, our partial evaluator reduces to specializing continuations and performing standard optimizations (like constant folding, algebraic simplifications, etc.). Unlike previous work [Leißa et al 2015a] on Thorin, our evaluator is entirely driven by implicit and explicit annotations (in the style of Schism filters [Consel 1988]). This makes PE predictable for the programmer because it eliminates the influence of the imprecision of static analyses that guide other evaluators (see Section 3.1).…”

Section: Fig 1 the Anydsl Systemmentioning

confidence: 99%

“…Jones and Glenstrup 2002;Jones et al 1993] suggest applying a termination analysis to conservatively ensure termination of PE. Leißa et al [2015a] propose to partially evaluate everything until the evaluator hits the first dynamic conditional and resume evaluation at the post-dominator. If further evaluation under this conditional is desired, a user annotation can restart evaluation there.…”

Section: Partial Evaluationmentioning

confidence: 99%

“…In this setting, the modified compiler recognizes domain-specific patterns and manipulates the program representation. An alternative is to use PE in order to remove the overhead of the shallow embedding [Hudak 1998;Leißa et al 2015a]. This way, the DSL implementer does not have to modify a compiler.…”

Section: Dsl Embeddingmentioning

confidence: 99%

See 2 more Smart Citations

AnyDSL: a partial evaluation framework for programming high-performance libraries

Leißa

Boesche

Hack

et al. 2018

Proc. ACM Program. Lang.

Self Cite

View full text Add to dashboard Cite

This paper advocates programming high-performance code using partial evaluation. We present a clean-slate programming system with a simple, annotation-based, online partial evaluator that operates on a CPS-style intermediate representation. Our system exposes code generation for accelerators (vectorization/parallelization for CPUs and GPUs) via compiler-known higher-order functions that can be subjected to partial evaluation. This way, generic implementations can be instantiated with target-specific code at compile time. In our experimental evaluation we present three extensive case studies from image processing, ray tracing, and genome sequence alignment. We demonstrate that using partial evaluation, we obtain high-performance implementations for CPUs and GPUs from one language and one code base in a generic way. The performance of our codes is mostly within 10%, often closer to the performance of multi man-year, industry-grade, manuallyoptimized expert codes that are considered to be among the top contenders in their fields. CCS Concepts: • Software and its engineering → Compilers; Parallel programming languages; • Hardware → Emerging languages and compilers; • Computing methodologies → Image processing; Ray tracing; • Applied computing → Bioinformatics;

show abstract

Section: Fig 1 the Anydsl Systemmentioning

confidence: 99%

Section: Partial Evaluationmentioning

confidence: 99%

See 1 more Smart Citation

AnyDSL: a partial evaluation framework for programming high-performance libraries

Leißa

Boesche

Hack

et al. 2018

Proc. ACM Program. Lang.

Self Cite

View full text Add to dashboard Cite

show abstract

“…x => stagedFoo(x) is not qualitatively better for the user than the quoted version. fixed [27]; i.e., which parts are known statically is fully determined in advance by the library designers. Future changes to enable more optimizations may break the library interface, as we describe further in Section 5.6, and any usages of the library in a slightly more dynamic setting are impossible.…”

Section: Previous Approachesmentioning

confidence: 99%

Quoted staged rewriting: a practical approach to library-defined optimizations

2017

View full text Add to dashboard Cite

Staging has proved a successful technique for programmatically removing code abstractions, thereby allowing for faster program execution while retaining a high-level interface for the programmer. Unfortunately, techniques based on staging suffer from a number of problems -ranging from practicalities to fundamental limitations -which have prevented their widespread adoption. We introduce Quoted Staged Rewriting (QSR), an approach that uses type-safe, pattern matchingenabled quasiquotes to define optimizations. The approach is "staged" in two ways: first, rewrite rules can execute arbitrary code during pattern matching and code reconstruction, leveraging the power and flexibility of staging; second, library designers can orchestrate the application of successive rewriting phases (stages). The advantages of using quasiquote-based rewriting are that library designers never have to deal directly with the intermediate representation (IR), and that it allows for non-intrusive optimizations -in contrast with staging, it is not necessary to adapt the entire library and user programs to accommodate optimizations.We show how Squid, a Scala macro-based framework, enables QSR and renders library-defined optimizations more practical than ever before: library designers write domainspecific optimizers that users invoke transparently on delimited portions of their code base. As a motivating example we describe an implementation of stream fusion (a well-known deforestation technique) that is both simpler and more powerful than the state of the art, and can readily be used by Scala programmers with no knowledge of metaprogramming.

show abstract

“…One of the most promising projects following this approach is AnyDSL [11]. It is designed similar to established solutions like language workbenches, which are discussed in detail in [12].…”

Section: Introduction and Related Workmentioning

confidence: 99%

Automatic Generation of Massively Parallel Codes from ExaSlang

Kuckuk

Köstler

2016

Computation

View full text Add to dashboard Cite

Domain-specific languages (DSLs) have the potential to provide an intuitive interface for specifying problems and solutions for domain experts. Based on this, code generation frameworks can produce compilable source code. However, apart from optimizing execution performance, parallelization is key for pushing the limits in problem size and an essential ingredient for exascale performance. We discuss necessary concepts for the introduction of such capabilities in code generators. In particular, those for partitioning the problem to be solved and accessing the partitioned data are elaborated. Furthermore, possible approaches to expose parallelism to users through a given DSL are discussed. Moreover, we present the implementation of these concepts in the ExaStencils framework. In its scope, a code generation framework for highly optimized and massively parallel geometric multigrid solvers is developed. It uses specifications from its multi-layered external DSL ExaSlang as input. Based on a general version for generating parallel code, we develop and implement widely applicable extensions and optimizations. Finally, a performance study of generated applications is conducted on the JuQueen supercomputer.

show abstract

Shallow embedding of DSLs via online partial evaluation

Cited by 13 publications

References 32 publications

AnyDSL: a partial evaluation framework for programming high-performance libraries

AnyDSL: a partial evaluation framework for programming high-performance libraries

Quoted staged rewriting: a practical approach to library-defined optimizations

Automatic Generation of Massively Parallel Codes from ExaSlang

Contact Info

Product

Resources

About