A prototype loop reverser for the scalarizer
Paul Thomas
paulthomas2@wanadoo.fr
Mon Oct 9 05:06:00 GMT 2006
Daniel and Steven,
Once again, thanks for the illuminating correspondence; you can put the
daggers away now!
>
> It is of course, up to whoever implements it to implement it where
> they want, as long as there are people willing to approve it. I can
> only offer an opinion as to where I would do it, and why
That is, of course, why I submitted the patch to the list and invited
comments as to whether or not I should proceed. It seems to me:
(i) That the reverser in the scalarizer is but a small thing - the need
is easily recognized in the existing dependency analysis and is simply
dealt wih by the scalarizer.
(ii) Loop interchange in the scalarizer ought to work - all the
requisite code is there; it just needs debugging.
(iii) A natural dividing line, between what is desirable to do in the
frontend and what should be left to the middle-end machinery, is that of
any comparative analysis of alternative branches.
(iv) I should find out more about the middle-end!
Best regards
Paul
More information about the Fortran
mailing list