[Developers] New script options

Larry Jacobson (NOAA Federal) larry.jacobson at noaa.gov
Mon Apr 1 05:14:20 PDT 2013


Arni-

1) Here are some results using the Microsoft compilerand University of 
Maine model on my humble desktop.  I paste them in below and attach a 
pdf as well.

2) Note that I was not able to compile with -g (I had to stop it after 
66 hrs).

3) Isill try to do this in Linux today (we have some servers that are 
supposed to be very fast).

4) If it helps, I can send you the model code and data but you will have 
to remember that I didn't write it...

Cheers!





On 3/28/2013 1:23 PM, Arni Magnusson wrote:
> Safe compilation will be the default in all scripts, and that default 
> will carry over to ADMB-IDE.
>
> The idea was to introduce a new option -o to explicitly build an 
> optimized model. So "admb -o simple" would do exactly what the scripts 
> do today, and "admb -s simple" would be equivalent to "admb simple".
>
> I don't have a model that takes long to build, but can someone 
> benchmark this:
>
>                       Build time | Executable size | Run time
>   admb bigmodel
>   admb -s bigmodel
>   admb -g bigmodel
>   admb -g -s bigmodel
>
> The results will help us decide how the compilation scripts should 
> work. For example, we could have safe compilation turn off the -O3 
> optimization - like Ian suggests. In this way, safe compilation would 
> be safe (checking array bounds) and also build fast. Sounds like a 
> reasonable default while developing a model.
>
> It's tempting to keep the script UI simple and couple together 
> functionality (bounds, optimization, libraries, debug) that would be 
> sensible to use together. I think this coupling should be based on the 
> results from the above benchmark.
>
> Arni
>
>
>
> On Thu, 28 Mar 2013, Ian Taylor - NOAA Federal wrote:
>
>> Hi Developers,
>>
>> I know it's not that helpful for folks like me to make suggestions 
>> that they're not skilled enough to implement, but how hard would it 
>> be for ADMB to add a -o option that would turn on or off the -O3 or 
>> -Ox? And wasn't -s going to become the default at some point, or was 
>> the only proposed change the one that Arni to set this as default in 
>> the IDE?
>>
>> -Ian
>>
>>
>>
>> On Thu, 28 Mar 2013, Larry Jacobson wrote:
>>
>>> Sorry, my original note summarizing approaches to reducing compile 
>>> time for ADMB using the MinGW or Visual C++ 2010 Express compilersin 
>>> Windows left out 3 important items.  Unfortunately, one cannot 
>>> recompile e-mails.
>>>
>>> In any case:
>>>
>>> 1) Optimization is turned off using the MinGW compiler by changing 
>>> line line sym=-O3 in adcomp.bat so that it reads sym=-O.
>>>
>>> 2) Optimization is turned off using the Microsoft compiler by 
>>> omitting -Ox from the first call to the compiler.  That is, use:
>>>
>>>> cl -c /EHsc -DUSE_LAPLACE -DWIN32 %opt% -D__MSVC32__=8 -I. 
>>>> -I"%ADMB_HOME%"\include -I"%ADMB_HOME%"\contrib %1.cpp
>>>>
>>> instead of:
>>>
>>>> cl -c /EHsc -DUSE_LAPLACE -DWIN32 %opt% /Ox -D__MSVC32__=8 -I. 
>>>> -I"%ADMB_HOME%"\include -I"%ADMB_HOME%"\contrib %1.cpp
>>>
>>> 3) The popular IDE for Windows uses the MinGW compilerwith a 
>>> slightly different adcomp.bat file that is probably in a directory 
>>> named something like C:\admb\admb101-gcc452-win32\**bin. To turn off 
>>> optimization in the IDE, change theline sym=-O3 in adcomp.bat so 
>>> that it reads sym=-O(same as #1).  See Arni's note belowand an 
>>> additional one from John Sibert below Arni's.
>>>
>>> Cheers!
>>>
>>>  Hi Larry,
>>

-- 
**********************
Larry Jacobson
National Marine Fisheries Service
Northeast Fisheries Science Center
166 Water Street
Woods Hole, MA 02543-1026
Voice: 508-495-2317
Fax: 508-495-2393
E-mail: larry.jacobson at noaa.gov
**********************

-------------- next part --------------
A non-text attachment was scrubbed...
Name: UniversityofMaineLobsterModelMicrosoftCompiler-32bit-LJacobson.pdf
Type: application/pdf
Size: 14167 bytes
Desc: not available
URL: <http://lists.admb-project.org/pipermail/developers/attachments/20130401/f5074105/attachment.pdf>


More information about the Developers mailing list