Introduction ::.
Current high-end applications usually exploit just a fraction of the theoretical performance of large platforms. Interactions among the hardware, system software, programming interface, and algorithm are extremely complex, and the implications for development of hybrid MPI+OpenMP Fortran/C/C++ applications are challenging. Typical application development and tuning scenarios involve the manual and separate use of compilers and performance tools, and program modifications based upon insights laboriously gleaned from their output.
The NSF ST-HEC-funded Copper project is a multi-university collaborative effort between the University of Houston, University of Tennessee-Knoxville, University of Illinois-Urbana Champaign (NCSA), and Virginia Tech. The project aims to raise the quality and efficiency of the application development and tuning process by building an integrated environment for program optimization that reduces the manual labor and guesswork of existing approaches. The corresponding interfaces of Copper enable the application developer, compiler and performance tools to collaborate to generate optimized code based upon a variety of sources of feedback, including performance data from "offline" and "online" development runs. The Copper environment combines robust, existing, open source software, a compiler, a program analysis tool and two performance tools with complementary features into a single, coherent system for collaborative application tuning. The NSF award abstract provides additional details about the Copper Project.
Project Goals::.
- Facilitate the development of hybrid MPI+OpenMP codes
- Establish an iterative approach to application tuning for high end architectures
- Reduce the human effort in application tuning
