Philippe Ranger wrote:

> 
> <<
> Consequently, it is not the best generalized
> approach to these types of loops.  Now since classes are actually
> pointers it is easy to see that the preponderance of arrays fall into
> this catagory, and so I believe that the compiler should handle them
> differently.  This was my "wish".
> >>
> 
> I suppose the array access by pointer incrementation is valid for While
> loops too, so you're not saying the For is less optimized than the While,
> but that in the former case there's an extra optimization available for the
> most common use, and it is not put in.

The while loop really isn't optimized at all in the way for loops are. 
A while loop is handled more like a bent if-then statement. 
Consequently, you can manually optimize access within the loop by proper
use of pointers, indicies etc. But you have to do it all yourself.  I
would prefer if the already complex for loop optimizations were made a
bit more complex and treated arrays of simple sized elements
differently.

> My view on it is that it is historical. TP was happy to use the cx Intel had

> provided for this (with the loop operator), and now that the optimizer is
> smarter by a degree of magnitude, somehow this former, trivial, optimization
> hasn't been removed to free a reg. Have I got this right?

Nope, like I said the ecx usage was coincidental.  Toss a routine call
into the loop and the counter will be shifted to a protected register
like esi.


-- 
Bob Lee
High Performance Delphi - http://www.econos.com/optimize/
Updated August 2