[Rd] [R] R and GNU gcc for OpenMP

[Prof Brian Ripley]

On Wed, 2 Sep 2009, Duncan Murdoch wrote:

> Debabrata Midya wrote:
>> Dear R users,
>>  Thanks in advance.
>> 
>> 
>> 
>> I am using R 2.9.1 on Windows XP.
>>  May I request you to assist me on the following:
>>  I have downloaded Rtools29.exe and installed it today 02 Sep 2009 and I 
>> have also provided the appropriate path variable for Rtools.
>>  I have checked the version of gcc compiler in Rtools.exe is 4.2.1.
>>  I have checked (in  http://gcc.gnu.org/) the latest available gcc compiler 
>> is 4.4.1. My objective is to use latest gnu gcc for openMP and I like to 
>> link it to R (by creating *.dll).
>>  How can I use latest version of gcc for openMP? And at the same time I 
>> like to use Rtools as well.
>> 
>
> I have moved this response to the R-devel mailing list, it's not really 
> appropriate for R-help.
>
> Generally speaking the compiler used for a .dll need not be the same one used 
> for the rest of an application, so R won't care if you work out how to use 
> gcc 4.4.1 to build your dll.  However:
>
> - The scripts for building packages are written and tested against gcc 
> 4.2.1; you may need to tweak them to work with a different compiler.
> - As far as I know, there's no gcc 4.4.1 available for Windows, but 4.4.0 is 
> available on the MinGW site.  So if you really need 4.4.1 you may need to 
> compile it yourself.
>
> So this is a pretty difficult project.   You can ask on the R-devel mailing 
> list for help, but there aren't many people who will be able to help you, so 
> you may be pretty much on your own.

To add to this:

- The MinGW gcc 4.4.0 is too buggy to build R successfully.

- I don't see that you need 4.4.0 to use OpenMP: I have used it with 
4.2.1. (And note: that is not an offer of help to do so.)

>
> Duncan Murdoch
>>  Once again, thank you very much for the time you have given.
>>  I am looking forward for your reply.
>>  Regards,
>>  Deb


-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595