Van Snyder wrote:
> Is it common for compilers to generate code that suppresses taking copies
> if the elements of the dummy argument are contiguous?
It is clear that the standard was made so that compilers would be smart enough
to do things like this and be able to pass contiguous sections efficiently
between assumed-shape and assumed-size, as well as be able to pass both
contiguous and non-contiguous sections efficiently as assumed-shape, with the
appropriate descriptors.
I know that more work is and needs to be done on this side of the optimization
of Fortran compilers. What worries me and annoys me is that these things are
not at all well documented by compiler vendors. Hey, what is the big secret
about what you do when you pass array sections? All that is doing is making
life more difficult for us the users since we now have to devise clever tests
to find out what it is you are doing.
Speaking of tests, does anybody have a set of tests/benchmarks for this issue?
I want to know the same answers as Van Snyder for LF95.
Thanks,
Aleksandar
--
_____________________________________________
Aleksandar Donev
[log in to unmask]
http://www.pa.msu.edu/~donev/
Department of Physics and Astronomy
Michigan State University
East Lansing, MI 48825
(517) 432-6770
_____________________________________________
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|