[Bug fortran/114626] New: Very long time for compilation the attached program, depends on value of a parameter

[Bug fortran/114626] New: Very long time for compilation the attached program, depends on value of a parameter

[arjen.markus895 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114626

            Bug ID: 114626
           Summary: Very long time for compilation the attached program,
                    depends on value of a parameter
           Product: gcc
           Version: 12.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: arjen.markus895 at gmail dot com
  Target Milestone: ---

Created attachment 57894
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=57894&action=edit
Source of the program causing the very long compile time

The attached program causes the gfortran compiler to take a very long time if
the parameter max_ij is set to 100. It may not even finish. If the parameter is
set to 13 (to at least give one result), it takes a few seconds but at least
finishes in a reasonable time.

This is with default compile options. I have not tried specific options.

(Note: it is a silly program, meant for demonstration only.)

[Bug fortran/114626] Very long time for compilation the attached program, depends on value of a parameter

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114626

anlauf at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           See Also|                            |https://gcc.gnu.org/bugzill
                   |                            |a/show_bug.cgi?id=97571
           Priority|P3                          |P4

--- Comment #1 from anlauf at gcc dot gnu.org ---
Note that

pack( [ (k, k = 1,max_sum)], &
[ ( count( [((i**3+j**3, i = 1,j), j = 1,max_ij)] == k ) > 1, k = 1,max_sum )]
)

is a constant expression that the frontend tries to simplify at compile time.
I assume that most of the time goes into handling of the array constructors.

There's no limit implemented that checks the complexity of such expressions
and to give up if that limit is reached.

An expression like the above could occur in a parameter definition, and a
limit then might terminate the compilation with an error.

Fortunately, you've already found a workaround.

For a related PR see pr97571.

[Bug fortran/114626] Very long time for compilation the attached program, depends on value of a parameter

[arjen.markus895 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114626

--- Comment #2 from Arjen Markus <arjen.markus895 at gmail dot com> ---
Thanks for the explanation - that is clear enough.

[Bug fortran/114626] Very long time for compilation the attached program, depends on value of a parameter

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114626

--- Comment #3 from anlauf at gcc dot gnu.org ---
Another option is to not declare max_sum as parameter but as variable.
With

    integer              :: max_sum = 2 * max_ij ** 3

the code compiles almost instantly, but it has noticeable runtime.

(But it performs much better here than with Intel... :)