Paul Millar wrote:
>
> On Wed, 20 Mar 2002, Allan Whiteford wrote:
> > Is their not a difference between optimising a piece of code (in a
> > textbook computer science type way) and choosing the correct algorithm
> > (by actually engaging your brain)?
>
> (IMHO) sort of ...
>
> > I'd say they were different things but I wouldn't like to have to draw a
> > line anywhere between them.
>
> Yes, I agree. In practise there is a difference, but conceptually there
> isn't.
>
I think there are conceptual differences.
As an example of a purely computational optimisation of a piece of code
consider finding the total of all the elements in a 10x10 array.
The algorithm would be as follows:
"Set the total to zero. Go through each of the elements in turn and add
it's value to the total".
2 possible implementations of this are:
total=0;
for (i=0;i<=9;i++)
{
for (j=0;j<=9;j++)
{
total=total+array[i][j];
}
}
total=0;
for (i=0;i<=9;i++)
{
for (j=0;j<=9;j++)
{
total=total+array[j][i];
}
}
The first solution will be faster but the algorithm hasn't changed. It's
just about knowing the language/compiler. However, once you go to
something complicated I think it's hard to label an optimisation as
being purely computational or something to do with the algorithm; it
just becomes a big grey area.
>
> Yep, the right tool for the right job. As long as you sort out the higher
> level stuff (which algorithm to use), let the compiler sort of some of the
> nitty gritty detail.
I agree completely. Nobody wants to unroll their own loops :).
Thanks,
Allan
--
Intel: We put the 'um...' in Pentium.
--------------------------------------------------------------------
http://www.lug.org.uk http://www.linuxportal.co.uk
http://www.linuxjob.co.uk http://www.linuxshop.co.uk
--------------------------------------------------------------------
