[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130
Bug 21130 depends on bug 25708, which changed state.

Bug 25708 Summary: [F95] Module loading is not good at all
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=25708

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
         Resolution|---                         |FIXED

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[paul dot thomas at jet dot uk]

------- Comment #16 from paul dot thomas at jet dot uk  2006-09-20 09:22 -------
(In reply to comment #15)
> looks like there is agreement that the problem is fixed.

I am afraid to say that I do not agree entirely; the above takes 6 seconds to
compile with ifort on my Athlon 1700 "a vapeur".  The patch is pretty good but
Andrew is right to say in PR25708 that we should not be loading the modules
repeatedly. I put a counter in read_module to check how many times each module
was read, in the above, and came to the conclusion that gfortran would achieve
the same compile time as ifort, if we were to read a module once into a
separate symtree and use and abuse it from there.

For my sins, I have proposed to do this as a gfortran-4.3 project.

Paul


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[bdavis at gcc dot gnu dot org]

------- Comment #15 from bdavis at gcc dot gnu dot org  2006-05-12 16:05 -------
looks like there is agreement that the problem is fixed.


-- 

bdavis at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[steven at gcc dot gnu dot org]

------- Comment #14 from steven at gcc dot gnu dot org  2006-05-05 18:42 -------
Bud already voted to close this in comment #11 :-)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[jvdelisle at gcc dot gnu dot org]

------- Comment #13 from jvdelisle at gcc dot gnu dot org  2006-05-05 05:40 -------
Bud, unless you are going to chase this further I vote that you close this PR. 
Was the patch going to go to 4.1 branch as well?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[bdavis at gcc dot gnu dot org]

-- 

bdavis at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |bdavis at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2005-09-19 00:49:39         |2006-04-17 22:53:54
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[bdavis at gcc dot gnu dot org]

------- Comment #12 from bdavis at gcc dot gnu dot org  2006-03-31 00:47 -------
Subject: Bug 21130

Author: bdavis
Date: Fri Mar 31 00:47:13 2006
New Revision: 112558

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=112558
Log:
2006-03-30  Paul Thomas <paulthomas2@wanadoo.fr>
            Bud Davis  <bdavis9659@sbcglobal.net>

        PR 21130
        * module.c (load_needed): Traverse entire tree before returning.



Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/module.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[bdavis at gcc dot gnu dot org]

------- Comment #11 from bdavis at gcc dot gnu dot org  2006-03-30 14:04 -------
if this patch gets accepted, i vote to close the PR.  we are still slower than
we should be, but it is now linear and compiles in a 'reasonable time'.


--bud


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[bdavis at gcc dot gnu dot org]

------- Comment #10 from bdavis at gcc dot gnu dot org  2006-03-30 13:58 -------
with the same patch, the aermod.f90 :

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total
 time   seconds   seconds    calls   s/call   s/call  name
  7.69      8.80     8.80 29723933     0.00     0.00  parse_atom
  5.26     14.82     6.02 125084260     0.00     0.00  module_char
  3.51     18.83     4.01                             record_reg_classes
  2.83     22.07     3.24                             find_reloads
  2.19     24.58     2.51                             walk_tree
  1.85     26.70     2.12  2331175     0.00     0.00  gt_ggc_mx_lang_tree_node
  1.77     28.73     2.03                             ht_lookup_with_hash
  1.72     30.70     1.97                             ggc_set_mark
  1.48     32.39     1.69                             ggc_alloc_stat
  1.37     33.96     1.57 10696036     0.00     0.00  set_module_locus
  1.27     35.41     1.45 13353395     0.00     0.00  require_atom
  1.14     36.71     1.30   722236     0.00     0.00  gfc_match_strings
  1.12     37.99     1.28   817282     0.00     0.00  insert
  1.12     39.27     1.28                             constrain_operands
  1.08     40.50     1.23 31922526     0.00     0.00  get_module_locus
  1.03     41.68     1.18     3220     0.00     0.00  traverse_ns
  0.94     42.76     1.08      706     0.00     0.02  load_needed


--bud


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[bdavis at gcc dot gnu dot org]

------- Comment #9 from bdavis at gcc dot gnu dot org  2006-03-30 13:52 -------
after the above patch, here is a profile of the last file (that takes so long
to compile):

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total
 time   seconds   seconds    calls   s/call   s/call  name
 15.52     19.04    19.04 60138476     0.00     0.00  parse_atom
  9.33     30.48    11.44 239125668     0.00     0.00  module_char
  5.22     36.88     6.40 89284617     0.00     0.00  compare_integers
  4.60     42.52     5.64                             ht_lookup_with_hash
  4.22     47.70     5.18  2224328     0.00     0.00  insert
  3.56     52.07     4.37      622     0.01     0.07  load_needed
  2.96     55.70     3.63  5543817     0.00     0.00  get_integer
  2.63     58.93     3.23  4035787     0.00     0.00  check_interface0
  2.32     61.78     2.85      312     0.01     0.03  read_cleanup
  2.31     64.61     2.83 74319506     0.00     0.00  get_module_locus
  2.28     67.41     2.80 21727690     0.00     0.00  set_module_locus
  1.77     69.58     2.17 25323260     0.00     0.00  require_atom
  1.59     71.53     1.95  4464297     0.00     0.00  clear_sym_mark
  1.57     73.45     1.92   720088     0.00     0.00  count_types_test
  1.56     75.36     1.91 13155211     0.00     0.00  compare_true_names
  1.50     77.20     1.84   948414     0.00     0.00  traverse_ns
  1.37     78.88     1.68  5424920     0.00     0.00  compare_type_rank
  1.37     80.56     1.68  5424108     0.00     0.00  compare_type_rank_if


--bud


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[howarth at nitro dot med dot uc dot edu]

------- Comment #8 from howarth at nitro dot med dot uc dot edu  2006-03-29 04:32 -------
I can confirm this patch resolves the test case on MacOS X with gcc 4.1
branch...

Before patch...

Tue Mar 28 23:06:11 EST 2006
The following file will take roughly 10 minutes to compile
Tue Mar 28 23:06:23 EST 2006
Tue Mar 28 23:25:00 EST 2006

After patch...

Tue Mar 28 23:26:00 EST 2006
The following file will take roughly 10 minutes to compile
Tue Mar 28 23:26:12 EST 2006
Tue Mar 28 23:27:03 EST 2006

This is on a dual 2.0GHz G5 with 2 Gb of RAM. However I don't see any
significant impact
on the compile time for the aermod polyhedron benchmark.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[bdavis at gcc dot gnu dot org]

------- Comment #7 from bdavis at gcc dot gnu dot org  2006-03-29 01:47 -------
proposed patch:

http://gcc.gnu.org/ml/fortran/2006-03/msg00517.html
http://gcc.gnu.org/ml/fortran/2006-03/msg00518.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[rguenth at gcc dot gnu dot org]

------- Comment #6 from rguenth at gcc dot gnu dot org  2006-03-02 16:10 -------
aermod of Polyhedron has the same problem.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-09-19 00:49 -------
Confirmed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-09-19 00:49:39
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[jvdelisle at verizon dot net]

------- Additional Comments From jvdelisle at verizon dot net  2005-04-24 04:05 -------
ifort does it in about 5 seconds.  It issues several warnings about alignments,
but otherwise completes quickly.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-04-21 00:14 -------
The compile time problem is located in load_needed in module.c but I don't know how to fix it easily.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-04-21 00:01 -------
Hmm, this comes from loading modules or something related to that.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |compile-time-hog


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130

[Bug fortran/21130] 38822 lines of Fortran 90 takes more than 10 minutes to compile on a dual 3GHz P4 Linux box with lots of RAM

[epperly2 at llnl dot gov]

------- Additional Comments From epperly2 at llnl dot gov  2005-04-20 22:46 -------
Created an attachment (id=8690)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=8690&action=view)
Source files and compile script to reproduce the performance issue

I am attaching a bzip2 compressed tar file with the source and a compile.sh
script to reproduce this issue.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21130