Hi.
(Xposted to both comp.lang.c++ and
comp.programming since I've got
questions related to both C++ language and general programming)
I've got the following C++ code. The first routine runs in like 65%% of
the time of the second routine. Yet both do the same thing. However,
the second one seems better in terms of the way the code is written
since it helps encapsulate the transformation in the inner loop better
making it easier to read, at least in my opinion. Maybe it's not,
maybe the former is "better" that way and I should go with it, but if
the latter is "better" in that respect should I just ditch it anyway
and tradeoff for performance since I want this thing to be fast???
What each routine does is multiply two arbitrary-precision integers
together. The second one though uses an additional "slice" type that
provides a window enabling the "multiply and add" operation to be
performed on a limited range of digits, which can then be advanced
across the number, making clear that part of the algorithn.
I'm using the simple "grade school" multiply algorithm. Note how the
second routine more easily outlines this algorithm, while in the first
it is a little more difficult to see. Which would you prefer, exactly?
