Next: : REFERENCES
Up: R. Clapp: Parallel datasets
Previous: FUTURE WORK
Developing code for a parallel environment
requires significant, sometimes non-obvious
additional coding overhead. By expanding the
concept of a SEPlib dataset to include a
dataset that is spread over several machines,
much of the difficulty associated with
cluster computing can be avoided by the
programmer. The included examples show
how with relatively minor changes in coding
style a parallel code can be created virtually
free.
Stanford Exploration Project
5/23/2004