Optimization of gfortran for execution speed

[Paul Thomas]

Dominique Dhumieres wrote:

>>...
>>- don't forget to check that you don't break stuff like
>>   a(1:N-1) = a(2:N)
>>...
>>    
>>
>
>That's the easy case without dependency, the dangerous one
>  
>
Damn, if you think that it is easy, would you please do it?  Trying to 
subtractions of expressions involving EXPR_CONSTANTs and EXPR_VARIABLEs 
has my head hurting.  I am sure that, as usual, the final result will 
look simple but getting there I am finding painful.

>is a(2:N) = a(1:N-1) that should be scalarized as
>a(N:2:-1) = a(N-1:1:-1) (I hope it's correct).
>  
>
Yes, I am aware of that one but.... one thing at a time!

Paul