[MITgcm-devel] Beaufort experiment on mac os x

[Menemenlis, Dimitris (3248)]

Martin, I change title for the Beaufort thread and cc Torge and Art

I'd love to get to bottom of this - it may be a bug in Beaufort set up
or a bug in latest Mac OS + gfortran compiler

My size is similar to yours:

bash-3.2$ size mitgcmuv 
__TEXT  __DATA  __OBJC  others  dec     hex
3080192 160526336       0       4295045120      4458651648      109c1a000

but when I actually run the code it only uses ~100 MB:

PID     COMMAND           %CPU   TIME      #TH   #WQ   #PORT #MREGS RPRVT   RSHRD   RSIZE   VPRVT   VSIZE   PGRP   PPID   STATE     UID   FAULTS     COW      MSGSENT     MSGRECV     SYSBSD     SYSMACH    CSW         PAGEINS  USER
45930   mitgcmuv          99.3   01:06.65  1/1   0     14    38     101M    240K    103M    172M    2538M   45930  45312  running   501   26524      44       89          44          105956+    66         6591+       0        dmenemen

I use an older version of gfortran:

bash-3.2$ gfortran -v
Using built-in specs.
Target: x86_64-apple-darwin10
Configured with: ../gcc-4.5-20090910/configure --enable-languages=fortran --host=x86_64-apple-darwin10 --build=x86_64-apple-darwin10
Thread model: posix
gcc version 4.5.0 20090910 (experimental) (GCC) 

I run on Mac OS-X Snow Leopard and I use:
# OPTFILE="../tools/build_options/darwin_ia32_gfortran"

I will compile code with a bunch of options that Art sent me to see if there are any funny loose ends.

Dimitris Menemenlis

On Mar 13, 2012, at 9:23 AM, Martin Losch wrote:

> Interesting. what does "size mitgcmuv" say?
> which compiler do you use? I have 
> run> size ../build/mitgcmuv
> __TEXT	__DATA	__OBJC	others	dec	hex
> 3407872	160079872	0	4295045120	4458532864	109bfd000
> so more than 4GB. Where does that come from? Version of gfortran?
> run> gfortran -v
> Using built-in specs.
> COLLECT_GCC=gfortran
> COLLECT_LTO_WRAPPER=/usr/local/libexec/gcc/x86_64-apple-darwin10.4.0/4.6.0/lto-wrapper
> Target: x86_64-apple-darwin10.4.0
> Configured with: ../gcc-4.6-20101106/configure --enable-languages=c++,fortran
> Thread model: posix
> gcc version 4.6.0 20101106 (experimental) (GCC) 
> 
> 
> M.
> 
> On Mar 13, 2012, at 4:53 PM, Menemenlis, Dimitris (3248) wrote:
> 
>> Gael and Jean-Michel, thanks very much for replies.
>> 
>> I am experimenting with updated sea ice code on my baby-Beaufort set-up:
>> http://mitgcm.org/viewvc/MITgcm/MITgcm_contrib/MPMice/beaufort/
>> 
>> Incidentally (for Torge, Martin, and Patrick), I can run multiple copies
>> of the Beaufort experiment on my 4-yr-old, 4-GB Macbook Pro.
>> Each copy requires ~100 MB of memory.  Strange that it would
>> run out of memory on newer Mac hardware?
>> 
>> Dimitris Menemenlis