I have been reading “Algorithms Illuminated” by Roughgarden, and I’ve been loving the book. However, it focuses on big-O run time analysis. And I’ve seen that data oriented design can dramatically change the run time of an algorithm. For example, processing so that the largest number of iterations can operate on data in cache can significantly help. (See Mike Acton’s talk on Data Oriented Design for details.) I know, for example, FFTW learns the hardware (with something called wisdom) to make its implementation faster.
I would LOVE to learn how to take advantage of cache and memory when implementing algorithms. Is there a good textbook for learning how to do so? If not, is there a good resource for learning how to do so?
Thank you!