Re: GNAT compiler switches and optimization

[Gautier <gautier@fakeaddress.nil>]

tkrauss wrote:

>    N := Positive'Value (Argument (1));
...
>    for I in A'range(1) loop
>       for J in A'range(2) loop
probably you have a penalty there against
>    do I = 1,N
>       do J = 1,N
I guess the GNAT compiler can't tell that due to
 >   A := new Real_Matrix(1..N, 1..N);
esp. because alloc. is dynamic, you indeed have:
   for I in 1..N loop
Then you should try that and even
   for I in reverse 1..N loop
Perhaps g95 does that optimization when Ada has to stick with the mentioned 
direction.
Also, compiling with -S and looking at the assembler code helps a lot...
HTH, Gautier
______________________________________________________________
Ada programming -- http://www.mysunrise.ch/users/gdm/gsoft.htm

NB: For a direct answer, e-mail address on the Web site!