Optimizing Data Structures in High-Level Programs: New Directions for Extensible Compilers based on Staging

High level data structures are a cornerstone of modern programming and at the same time stand in the way of compiler optimizations. In order to reason about user or library defined data structures compilers need to be extensible. Common mechanisms to extend compilers fall into two categories. Frontend macros, staging or partial evaluation systems can be used to programmatically remove abstraction and specialize programs before they enter the compiler. Alternatively, some compilers allow extending the internal workings by adding new transformation passes at different points in the compile chain or adding new intermediate representation (IR) types. None of these mechanisms alone is sufficient to handle the challenges posed by high level data structures. This paper shows a novel way to combine them to yield benefits that are greater than the sum of the parts. Instead of using staging merely as a front end, we implement internal compiler passes using staging as well. These internal passes delegate back to program execution to construct the transformed IR. Staging is known to simplify program generation, and in the same way it can simplify program transformation. Defining a transformation as a staged IR interpreter is simpler than implementing a low-level IR to IR transformer. With custom IR nodes, many optimizations that are expressed as rewritings from IR nodes to staged program fragments can be combined into a single pass, mitigating phase ordering problems. Speculative rewriting can preserve optimistic assumptions around loops. We demonstrate several powerful program optimizations using this architecture that are particularly geared towards data structures: a novel loop fusion and deforestation algorithm, array of struct to struct of array conversion, object flattening, and code generation for heterogeneous parallel devices. We validate our approach using several non trivial case studies that exhibit good speedups in experiments.

Optimizing Data Structures in High-Level Programs New Directions for Extensible Compilers based on Staging

Nada Amin, Manohar Jonnalagedda, Vojin Jovanovic, Martin Odersky, Tiark Rompf

High level data structures are a cornerstone of modern programming and at the same time stand in the way of compiler optimizations. In order to reason about user or library-defined data structures, compilers need to be extensible. Common mechanisms to extend compilers fall into two categories. Frontend macros, staging or partial evaluation systems can be used to programmatically remove abstraction and specialize programs before they enter the compiler. Alternatively, some compilers allow extending the internal workings by adding new transformation passes at different points in the compile chain or adding new intermediate representation (IR) types. None of these mechanisms alone is sufficient to handle the challenges posed by high level data structures. This paper shows a novel way to combine them to yield benefits that are greater than the sum of the parts. Instead of using staging merely as a front end, we implement internal compiler passes using staging as well. These internal passes delegate back to program execution to construct the transformed IR. Staging is known to simplify program generation, and in the same way it can simplify program transformation. Defining a transformation as a staged IR interpreter is simpler than implementing a low-level IR to IR transformer. With custom IR nodes, many optimizations that are expressed as rewritings from IR nodes to staged program fragments can be combined into a single pass, mitigating phase ordering problems. Speculative rewriting can preserve optimistic assumptions around loops. We demonstrate several powerful program optimizations using this architecture that are particularly geared towards data structures: a novel loop fusion and deforestation algorithm, array of struct to struct of array conversion, object flattening and code generation for heterogeneous parallel devices. We validate our approach using several non trivial case studies that exhibit order of magnitude speedups in experiments.

Assoc Computing Machinery2013

Lightweight Modular Staging and Embedded Compilers

Tiark Rompf

Programs expressed in a high-level programming language need to be translated to a low-level machine dialect for execution. This translation is usually accomplished by a compiler, which is able to translate any legal program to equivalent low-level code. But for individual source programs, automatic translation does not always deliver good results: Software engineering practice demands generalization and abstraction, whereas high performance demands specialization and concretization. These goals are at odds, and compilers can only rarely translate expressive high-level programs tomodern hardware platforms in a way that makes best use of the available resources. Explicit program generation is a promising alternative to fully automatic translation. Instead of writing down the program and relying on a compiler for translation, developers write a program generator, which produces a specialized, efficient, low-level program as its output. However, developing high-quality program generators requires a very large effort that is often hard to amortize. In this thesis, we propose a hybrid design: Integrate compilers into programs so that programs can take control of the translation process, but rely on libraries of common compiler functionality for help. We present Lightweight Modular Staging (LMS), a generative programming approach that lowers the development effort significantly. LMS combines program generator logic with the generated code in a single program, using only types to distinguish the two stages of execution. Through extensive use of component technology, LMS makes a reusable and extensible compiler framework available at the library level, allowing programmers to tightly integrate domain-specific abstractions and optimizations into the generation process, with common generic optimizations provided by the framework. Compared to previous work on programgeneration, a key aspect of our design is the use of staging not only as a front-end, but also as a way to implement internal compiler passes and optimizations, many of which can be combined into powerful joint simplification passes. LMS is well suited to develop embedded domain specific languages (DSLs) and has been used to develop powerful performance-oriented DSLs for demanding domains such as machine learning, with code generation for heterogeneous platforms including GPUs. LMS has also been used to generate SQL for embedded database queries and JavaScript for web applications.

EPFL2012

Spiral in Scala: Towards the Systematic Construction of Generators for Performance Libraries

Martin Odersky, Tiark Rompf, Alen Stojanov

Program generators for high performance libraries are an appealing solution to the recurring problem of porting and optimizing code with every new processor generation, but only few such generators exist to date. This is due to not only the difficulty of the design, but also of the actual implementation, which often results in an ad-hoc collection of standalone programs and scripts that are hard to extend, maintain, or reuse. In this paper we ask whether and which programming language concepts and features are needed to enable a more systematic construction of such generators. The systematic approach we advocate extrapolates from existing generators: a) describing the problem and algorithmic knowledge using one, or several, domain-specific languages (DSLs), b) expressing optimizations and choices as rewrite rules on DSL programs, c) designing data structures that can be configured to control the type of code that is generated and the data representation used, and d) using autotuning to select the best-performing alternative. As a case study, we implement a small, but representative subset of Spiral in Scala using the Lightweight Modular Staging (LMS) framework. The first main contribution of this paper is the realization of c) using type classes to abstract over staging decisions, i.e. which pieces of a computation are performed immediately and for which pieces code is generated. Specifically, we abstract over different complex data representations jointly with different code representations including generating loops versus unrolled code with scalar replacement-a crucial and usually tedious performance transformation. The second main contribution is to provide full support for a) and d) within the LMS framework: we extend LMS to support translation between different DSLs and autotuning through search.