CRPC-TR94378-S

Title: Requirements for Data-Parallel Programming Environments


Authors: Vikram Adve, Alan Carle, Elana Granston, Seema Hiranandani, 
         Ken Kennedy, Charles Koelbel, Ulrich Kremer, John Mellor-
         Crummey, Chau-Wen Tseng, and Scott Warren

Of the authors listed above, the following are

   Minority authors: 

   Female authors: Elana Granston, Seema Hiranandani

   Student authors: Ulrich Kremer


Date: April 1994


Keywords (list up to 8): Data Parallelism, Programming Environments,
Source-Level Debugging, Static Performance Analysis, Dynamic Performance
Analysis, Data Mapping Assistance, Interfaces


Abstract:

   In this paper, we present our vision for the functionality and
organization of a data-parallel programming environment. We believe that
the primary goal of, as well as the primary challenge for, such an
environment is to support the development of efficient data-parallel
programs while insulating the programmer from the intricacies of the
explicitly parallel code. Such support is essential because a data-parallel
program will be extensively transformed during compilation into an
optimized, explicitly parallel program. Understanding these transformations
along with the architecture-specific details of parallelism is an
unnecessarily onerous requirement on the programmer. The central purpose of
portable, data-parallel languages like HPF is to eliminate this burden.


Publication History:

   Published in: IEEE Parallel and Distributed Technology, Systems 
                 and Applications, Fall 1994, pp.48-58.