Hi,

My supervisor requested that I should specify the theoretical scalability of processing steps using the Big-O notation: O(W(n)/p+T(n)) where W(n) is the parallelizeable part and T(n) is the serializeable part of a processing step.

I have to admit, although I've searched in various books for this definition, I've never seen scalability defined by using the Big-O notation.

And I also don't get the idea. An O(N^2) method will stay O(n^2) regardless how many CPUs are used for processing.

I am quite confused and would be really grateful for some enlightenmend.

Regards, Alexander