Windows With C++: High-Performance Algorithms

Algorithm efficiency is not as straightforward as you might think. A well-designed algorithm on a single processor can often outperform an inefficient implementation on multiple processors.

In the October 2008 issue of MSDN Magazine, Kenny Kerr walks through the development of a fairly simple algorithm with efficiency and scalability in mind.

Check the magazine archives for more on programming for performance and previous installments of the Windows With C++ column.

Comments (1)

  1. Michiel says:

    It’s still hard to take Microsoft seriously, if they still think this article represents "high-performance" algorithms. Summary so you don’t have to read it: use your CPUs and access your memory row-first for memory locality. Heck, in the second minute of any image processing algorithms course, they’ll point out that staggering rows may offer an even greater performance. And the course might also mention the MMX extensions, introduced in 1997.