[MITgcm-devel] Beaufort experiment on mac os x
Torge Martin
torge.martin at gmail.com
Thu Mar 15 13:56:16 EDT 2012
I only get beaufort compiled with -O0. Too bad -- but at least its
running, finally.
Maybe I should dig out an older gfortran version ...
Torge
On 3/15/12 10:27 AM, Menemenlis, Dimitris (3248) wrote:
> Martin, from Torge's message:
>> same error occurs with -O2
> So it looks he may need to reduce optimization level even more.
>
> Torge, could you try to add to NOOPTFILE-list
> as per Martin's suggestion, or use "-O0".
>
> I just updated my own copy of code, to get
> Jean-Michel's latest gmredi_calc_psi_bvp.F
> and I compile/run fine on my older Mac with
> older compiler version and standard "-O3" optfile.
>
> Dimitris Menemenlis
>
> On Mar 15, 2012, at 10:21 AM, Martin Losch wrote:
>
>> It looks again like an optimization problem within the compiler.
>>
>> Jean-Michel just changed gmredi_calc_psi_bvp.F so maybe you want to put that into the NOOPTFILE-list as well, although I think we should revisit why it has problems after the update.
>>
>> The alternative is (as usual) to reduce to -O2 in general, because gfortran is obviously not bug-free (it's open-source free software, so I don't blame them). You could summit a bugreport (to bugzilla).
>>
>> Martin
>>
--
----------------------------------------------
Dr. Torge Martin
Meteorologist (Dipl.-Met.) / Sea Ice Physicist
Postdoctoral Scientist at
Massachusetts Institute of Technology
Earth, Atmospheric and Planetary Sciences
currently visiting the
Polar Science Center / Applied Physics Lab.
University of Washington
1013 NE 40th St
Seattle, WA 98105
USA
phone: +1 206 685-2262
fax: +1 206 616-3142
torge.martin at gmail.com
http://sites.google.com/site/torgemartin
----------------------------------------------
More information about the MITgcm-devel
mailing list