This week Daniel Moth and I have an article in Dr. Dobbs, Microsoft's C++ AMP Unveiled, which provides a fairly condensed introduction (or maybe more aptly we should just call it a teaser) to the technology. Articles such as the one we just wrote are limited to a certain number of words, so at parts it may appear steep and terse. Therefore, if you have any questions or need clarification, please do post them here and we will do our best to address them.
In the first part of the article, we go through a whirlwind tour of the programming model (which is only possible because the programming model is really thin and light), and then we present a couple of C++ AMP examples, which calculate a simple moving average of a number series using two different techniques and levels of sophistication. This will serve as the basis for future articles or blog posts which will continue to look at this problem from different angles, outlining elements in the programming model as well as the performance characteristics of GPU execution.