Embedded Answers

Inspiration

Rick Beato and Scott Devine both struggled to get into music school. Now they have, in their own very different ways, created an enormous wealth of freely available information, in part to help people who cannot get access to music school.

With Embedded Answers, I would like to make some of the information that you might get in a good computer science degree available to people who cannot easily get to college. Or they went to college and this material was not covered.

Richard Bird

Richard Bird gave a seminar in the mid 90s, together with research collaborator Ooge de Moor. It changed, forever, my understaning of high-performance computing. The academic view of computer performance (that I am steriotyping, here) is that only the algorithm matters. It is true that no amount of hacking around with assembly code will make the wrong algorithm perform well, and that the algorithm choice is vital for large data sets.</p>

Richard Bird was the first academic, in my experience, who came out and said that you need both. Once you have the right algorithm, you can still double the performance with low-level optimization. And doubling the performance matters.

I have since coined the terms macro-optimization for choice of algorithm and data-structures and micro-optimization for tuning at the instruction level. It is not a perfect classification, but I find it helpful to have a way to know which kind of optimization we are thinking about.

Mediumish

This website is built with Jekyll and Mediumish template for Jekyll. The Jekyll template is free to use. It is requested that users donate and acknowledge WowThemes.net.