Annotating user-defined abstractions for optimization
This paper discusses the features of an annotation language that we believe to be essential for optimizing user-defined abstractions. These features should capture semantics of function, data, and object-oriented abstractions, express abstraction equivalence (e.g., a class represents an array abstraction), and permit extension of traditional compiler optimizations to user-defined abstractions. Our future work will include developing a comprehensive annotation language for describing the semantics of general object-oriented abstractions, as well as automatically verifying and inferring the annotated semantics.
- Research Organization:
- Lawrence Livermore National Laboratory (LLNL), Livermore, CA
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 889969
- Report Number(s):
- UCRL-PROC-217573
- Country of Publication:
- United States
- Language:
- English
Similar Records
A Source-to-Source Architecture for User-Defined Optimizations
Applying Loop Optimizations to Object-oriented Abstractions Through General Classification of Array Semantics
The Specification of Source-to-source Transformations for the Compile-time Optimization of Parallel Object-oriented Scientific Applications
Conference
·
Wed Feb 05 23:00:00 EST 2003
·
OSTI ID:15004527
Applying Loop Optimizations to Object-oriented Abstractions Through General Classification of Array Semantics
Conference
·
Thu Mar 04 23:00:00 EST 2004
·
OSTI ID:15014021
The Specification of Source-to-source Transformations for the Compile-time Optimization of Parallel Object-oriented Scientific Applications
Conference
·
Tue Jun 05 00:00:00 EDT 2001
·
OSTI ID:15006272