Software Technology

Development Tools

Autotuning Compiler Technology for Cross-Architecture Transformation and Code Generation

Principal Investigators: Mary Hall, lead, University of Utah; Sam Williams, Lawrence Berkeley National Laboratory; Paul Hovland, Argonne National Laboratory

The goal of this project is to enable application developers to write and maintain a single copy of their source code, legal code that can be compiled and run by using standard tools, and (semi-)automatically map to a variety of high-performance, architecture-specific implementations. The approach is to separate a high-level C/C++/FORTRAN implementation from architecture-specific implementation (OpenMP, CUDA, etc.), optimization, and tuning. The autotuning compiler and search framework will transform the baseline code into a collection of highly-optimized implementations. Thus, autotuning will mitigate the need for extensive manual tuning.