Reusable, generic program analyses and transformations

Abstract

The optimizations in modern compilers are constructed for a predetermined set of primitive types. As a result, programmers are unable to exploit optimizations for user-defined types where these optimizations would be correct and beneficial. Moreover, because the set of optimizations is also fixed, programmers are unable to incorporate new optimizations into the compiler. To address these limitations, we apply the reuse methodologies from generic programming to compiler analyses and optimizations. To enable compilers to apply optimizations to classes of types rather than particular types, we define optimizations in terms of generic interface descriptions (similar to C++ concepts or Haskell type classes). By extending these interface descriptions to include associated program analysis and transformation fragments, we enable compilers to incorporate user-defined transformations and analyses. Since these transformations are explicitly associated with interface descriptions, they can be applied in generic fashion by the compiler. We demonstrate that classical compiler optimizations, when generalized using this framework, can apply to a broad range of types, both built-in and user-defined. Finally, we present an initial implementation, the principles of which are generalizable to other compilers.

Does Code Generation Promote or Prevent Optimizations?

“…Then, they optimize the CFG produced from the AST. Based on this modified CFG, generic optimizations are written in the specification language Scheme [18]. Although this approach, like our approach, is not satisfied with modern compiler optimizations and intends to enhance them, we disapprove the idea of qualifying the compiler optimization set as fixed set.…”

“…There are other approaches that are not satisfied by compiler optimizations such as [17] and [18]. In [18], authors argue that optimizations in modern compilers are constructed for a predetermined set of primitive types.…”

“…In [18], authors argue that optimizations in modern compilers are constructed for a predetermined set of primitive types. Thus, programmers are unable to exploit optimizations for user-defined types.…”

Exploiting domain knowledge to optimize parallel computational mechanics codes

“…Even though the transformations are generic, they can be driven by domainspecific cost models as in our application study, effectively performing domain-specific optimizations while remaining domain agnostic. This is similar in spirit to the approach of Willcock et al [28], where compiler transformations that deal with basic types are extended to work on more complex types, allowing for the same transformation pattern to be used in multiple domains; the key difference is that in our scenario, transformations are driven by domain-specific cost models, rather than syntax.…”

Programming for Exascale Computers

“…DSLs and libraries are particular examples of "multilevel programming": The application scientist programs in a notation that is meaningful to her, while the performance programmer that implements the libraries programs at a level that is closer to the hardware. Mechanisms and tools that facilitate such a division of labor are a subject of intensive research [53], [54].…”