[Bug libfortran/78549] [8 Regression] Very slow formatted internal file output

[tkoenig at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78549

Thomas Koenig <tkoenig at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW
   Target Milestone|---                         |8.0
            Summary|Very slow formatted         |[8 Regression] Very slow
                   |internal file output        |formatted internal file
                   |                            |output

--- Comment #13 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
Here is the timing for gcc 6:

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total           
 time   seconds   seconds    calls  ms/call  ms/call  name    
 32.15      0.99     0.99                             output_float
 12.34      1.37     0.38                             formatted_transfer
 11.04      1.71     0.34                             next_format0
  7.14      1.93     0.22                             write_float
  3.73      2.05     0.12                             _gfortrani_write_block
  3.25      2.15     0.10                             next_char
  2.92      2.24     0.09                             data_transfer_init
  2.60      2.32     0.08                             _gfortrani_next_format
  2.60      2.40     0.08                             get_fnode
  2.60      2.48     0.08                             write_integer
  2.27      2.55     0.07                             _gfortrani_parse_format
  2.27      2.62     0.07                             finalize_transfer
  2.27      2.69     0.07                             parse_format_list
  1.95      2.75     0.06                             _gfortrani_next_record
  1.14      2.78     0.04                             _gfortran_transfer_array
  0.97      2.81     0.03                             _gfortrani_mem_alloc_r4

So, this does appear to be a regression for gcc8 at least.

The time is also longer (~20 s for trunk, ~14 s for gcc 6).