Algorithms and Parallel Computing

1.11 APPLICATIONS OF PARALLEL COMPUTING

The availability of inexpensive yet really powerful parallel computers is expected to make a hitherto unforeseeable impact on our lives. We are used now to parallel computers helping us access any information through web search engines. In fact, the search progresses as we are typing our search key words. However, there is room for improvement and, more importantly, for innovation, as the following sections illustrate.

1.11.1 Climate Modeling

Climate simulations are used for weather forecasting as well as for predicting global climate changes based on different phenomena or human activities. As Reference 1 points out, the resolution of today’s climate models is 200 km. This is considered low resolution given the fact that some climate systems exist completely within such resolution scale.

Assume a high-resolution model for climate simulation partitions the globe using 3-D cells 1 km in size in each direction. Assume also that the total surface of the earth to be 510 × 10⁶ km² and the thickness of the atmospheric layer to be approximately 1,000 km. Then, we need to simulate approximately 5 × 10¹¹ weather cells. Assume further that each cell needs to do 200 floating point operations for each iteration of the simulation. Thus, we have to perform a total of 10¹⁴ floating point operations per iteration.

Let us now assume that we need to run the simulation 10⁶ times to simulate the climate over some long duration of the weather cycle. Thus, we ...