Parallel
Two Variations of Parallel Pipelines
To programmers unfamiliar with parallel pipelines the parallelization strategy tends to be: Use a profiler to identify hot spots and address your parallelization efforts to those areas first. Repeat profiling and parallelizing hot spots until diminishing returns indicate your parallelization task is complete. The next step would be to look at improving the I/O by adding multi-buffering for input and output as well as adding additional threads and program complexity to perform the I/O. Another description of this process is: as your returns diminish, increase the complexity of your program. The programmers familiar with parallel pipelines will recognize the above strategy has the priorities reversed, often produces non-optimal code, and consumes unnecessary programming efforts. To illustrate this point consult the following chart:Related Reading
More Insights
INFO-LINK
To upload an avatar photo, first complete your Disqus profile. | View the list of supported HTML tags you can use to style comments. | Please read our commenting policy. | |