An approach to ordering optimizing transformations
PPOPP '90 Proceedings of the second ACM SIGPLAN symposium on Principles & practice of parallel programming
BURG: fast optimal instruction selection and tree parsing
ACM SIGPLAN Notices
Automatic construction of optimizing parallelizing compilers from specification
Automatic construction of optimizing parallelizing compilers from specification
The Design of the PROMIS Compiler—Towards Multi-Level Parallelization
International Journal of Parallel Programming - Special issue on international symposium on high performance computing 1997, part I
Designing the McCAT Compiler Based on a Family of Structured Intermediate Representations
Proceedings of the 5th International Workshop on Languages and Compilers for Parallel Computing
The Design of the PROMIS Compiler
CC '99 Proceedings of the 8th International Conference on Compiler Construction, Held as Part of the European Joint Conferences on the Theory and Practice of Software, ETAPS'99
The SUIF Compiler System: a Parallelizing and Optimizing Research Compiler
The SUIF Compiler System: a Parallelizing and Optimizing Research Compiler
Effective Cross-Platform, Multilevel Parallelism via Dynamic Adaptive Execution
IPDPS '02 Proceedings of the 16th International Parallel and Distributed Processing Symposium
| Hi-index | 0.00 |
Implementing an optimizing compiler for a new target architecture has traditionally been a complex design/development effort requiring a large time scale. Existing machine descriptions and approaches based on pre-existing internal representations (IR) are not sufficient to build truly modular and extensible compilers. This paper describes the features of the Extensible Compiler Interface (ECI) implemented in the PROMIS compiler, which tackles several major problems concerning the reuse of compiler components, retargeting as well as extending existing compilers with new functionality. One of the main design issues is maintaining analysis information calculated by one module after another potentially unknown module modifies the IR. Another problem is expanding existing modules (or passes) to work with processor-specific instructions and data types added by the compiler developers. Our approach to compiler extensibility through the proposed ECI tackles and solves the above problems, and provides a simple yet powerful API for adding arbitrary functionality or entirely new optimizations to existing compilers. A case study is presented in which the components of a parallelizing compiler are reused to build a compiler for a vector architecture, thereby demonstrating the utility and convenience of ECI.