Feldspar: A domain specific language for digital signal processing algorithms

Abstract

Abstract-A new language, Feldspar, is presented, enabling high-level and platform-independent description of digital signal processing (DSP) algorithms. Feldspar is a pure functional language embedded in Haskell. It offers a high-level dataflow style of programming, as well as a more mathematical style based on vector indices. The key to generating efficient code from such descriptions is a high-level optimization technique called vector fusion. Feldspar is based on a low-level, functional core language which has a relatively small semantic gap to machine-oriented languages like C. The core language serves as the interface to the back-end code generator, which produces C. For very small examples, the generated code performs comparably to hand-written C code when run on a DSP target. While initial results are promising, to achieve good performance on larger examples, issues related to memory access patterns and array copying will have to be addressed.

Squid: type-safe, hygienic, and reusable quasiquotes

“…Earlier approaches such as LINQ [Meijer et al 2006] also provided some level of language-integrated domain-specific program reification. [Najd et al 2016] use TTH to build DSL programs for their alternative embedding of Feldspar [Axelsson et al 2010], an approach they call Quoted DSLs (QDSL). In this approach, a particular DSL is implemented using the quasiquotation abilities of a host language, which requires significant heavy lifting behind the scenes (for example, retrieving type information [Najd et al 2016]).…”

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

“…A major limitation of MSP is that it generally offers no type-safe facilities to analyse code (it is purely generative), which greatly restricts its capabilities in terms of program optimization. Moreover, staging a library exposes users of that library to staging annotations, 1 which leak through its interface. Perhaps more importantly, staging requires to decide from the beginning which parts of the program are static (meant to be executed at program generation time) and which parts are dynamic, then building everything around that dichotomy, making it hard to evolve the design later on without extensive refactorings.…”

“…Staging has been successfully applied to optimizing domain specific languages (DSL), especially Embedded DSLs (EDSL) [17] which are DSLs that are defined within a more expressive host language such as Haskell [1,17,29] or Scala [26,30,35,39]. In this context, staging has been used to facilitate the definition of extensible compilers for performanceoriented DSLs and heterogeneous target platforms [1,9,26,30,33].…”

“…In this context, staging has been used to facilitate the definition of extensible compilers for performanceoriented DSLs and heterogeneous target platforms [1,9,26,30,33]. Generally speaking, these compilers reuse the frontend capabilities of their host (syntax and type system) but they convert programs into their own domain-specific intermediate representation (IR) before stringifying low-level code.…”

Using C language extensions for developing embedded software: a case study

“…Language workbenches [13,14] are tools for efficiently implementing languages and their integrated development environments (IDEs). The JetBrains Meta Programming System (MPS) 1 is an open-source language workbench with comprehensive support for specifying structure, syntax, type systems, transformations and generation, debuggers and IDE support (see Figure 2). MPS relies on a projectional editor.…”