University of Illinois project on language analysis for Lawrence Livermore National Laboratory. Final report
The importance of parallel processing is a well-known fact. The limitations of parallel processing as practiced today are also well known. These limitations appear in the areas of software and computer organization. Many of the software problems studied nowadays are common to parallel and sequential processing. The problem of software portability is, however, particularly important for parallel processing. This problem arises from the need to convert lengthy pieces of code written for sequential processing into parallel programs. Also, the organization of parallel computers keeps evolving, and programs written for a particular parallel computer organization (e.g., a pipeline computer) may not be in the most suitable form for execution in another (e.g., a multiprocessor). On the other hand, the organization of present-day parallel computers seriously limits the speed of many programs. For example, pipeline computers are only appropriate for programs that either have a high density of vector operations, or can be transformed into programs with such high density. The research reported here studied ways to overcome both limitations.
- Research Organization:
- Lawrence Livermore National Lab., CA (USA); Illinois Univ., Urbana (USA). Dept. of Computer Science
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 6061832
- Report Number(s):
- UCRL-15412; ON: DE82003162
- Country of Publication:
- United States
- Language:
- English
Similar Records
Data Locality Enhancement of Dynamic Simulations for Exascale Computing (Final Report)
DHS-STEM Internship at Lawrence Livermore National Laboratory