Question

I have read a lot about Cache Oblivious Algorithms and Streaming trees etc. I understand the basics what I am still unable to see is why they are good for parallel programming? I think I saw John Harrop stated that they are revolutionary for that.

Answer 1

In the article
http://www.1024cores.net/home/parallel-computing/cache-oblivious-algorithms

They point out that

The idea behind cache-oblivious algorithms is efficient usage of processor caches and reduction of memory bandwidth requirements. Both things are equally important for single-threaded algorithms, but especially crucial for parallel algorithms, because available memory bandwidth is usually shared between hardware threads and frequently becomes a bottleneck for scalability.

Access to memory can be a bottle neck in parallel algorithms, so having algorithms that attempt to utilize cached memory can be more efficient.

In the same article they go on to describe how cache oblivious algorithms take advantage of the cache available:

Cache-oblivious algorithms work by recursively dividing a problem’s dataset into smaller parts and then doing as much computations of each part as possible. Eventually subproblem dataset fits into cache, and we can do significant amount of computations on it without accessing memory