[gmx-users] FFTW libraries

[Erik Lindahl]

Hi Seth,

I would suggest rebuilding a gromacs RPM for your system from the RPM 
sourcefile. It's really easy:

1. Install the rpm source gromacs-3.2.x-1.src.rpm (or something like 
that)
2. Change directory to /usr/src/redhat/SPECS
3. Build the binary packages: rpmbuild -bb gromacs.spec
4. Your custom rpm is now located in /usr/src/redhat/RPMS/i686.
5. Install it.

Cheers,

Erik

On Jun 22, 2004, at 6:36 AM, Dr Seth Carlton OLSEN wrote:

> Hi GROMACS users,
>
> Thank you, Choon-Peng, for explaining the difference between the 's' 
> and 'd' libraries for FFTW.  I have followed your suggestions and 
> those on the GROMACS site using:
> configure --enable-float --enable-type-prefix
> make
> make install
>
> followed by the double-precision installation (i.e. no enable-float 
> option)
>
> This seems to have put the following files in my /usr/local/lib 
> directory:
> libdfftw.a   libdrfftw.a   libsfftw.a   libsrfftw.a
> libdfftw.la  libdrfftw.la  libsfftw.la  libsrfftw.la
>
> However, I cannot find libsfftw.so.2 nor libsrfftw.so.2 in 
> /usr/local/lib or /usr/lib , and the command
> find / -name *fftw*so.2 -print
> yields no results.  Is it possible that when 'making' the program, 
> these libraries were skipped due to the prescence of more recent 
> libraries without the type prefixes enabled?
>
> Cheers,
>
> Seth
>
>
>
> ccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccms
>
> Dr Seth Olsen, PhD
> Postdoctoral Fellow, Computational Systems Biolgy Group
> Centre for Computational Molecular Science
> Chemistry Building,
> The University of Queensland
> Qld 4072, Brisbane, Australia
>
> tel (617) 33653732
> fax (617) 33654623
> email: s.olsen1 at uq.edu.au
> Web: www.ccms.uq.edu.au
>
> ccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccms
>
>
>
>
>
> ----- Original Message -----
> From: Chng Choon Peng <cpchng at bii.a-star.edu.sg>
> Date: Tuesday, June 22, 2004 2:09 pm
> Subject: Re: [gmx-users] FFTW libraries
>
>> Hi Seth,
>>
>>  If I remember correctly, libfftw is for complex to complex
>> transforms and
>> librfftw is for real to complex (forward FFT).
>> The 's' is for single-precision, which is the default installation
>> mode for Gromacs on SSE/2 machines.
>> So, your 'hack' works if your installed FFTw is in single-precision
>> mode,which may not be so.
>>
>> When compiling FFTw from source, you can specify the flag
>> "--enable-type-prefix" and you'll get libsfftw and libdfftw for
>> single- and
>> double-precision transforms.
>> Probably you can find RPMs for each version, with the prefix.
>>
>>
>> cheers,
>> Choon-Peng
>> -- 
>> Mr. Choon-Peng CHNG
>> Research Associate
>> Computational Biology Group
>> BioInformatics Institute, BMSI, A*STAR
>> 30 Biopolis Street
>> #07-01 Matrix Building
>> Singapore 138671
>> Tel (O): +65 64788301 Fax (O): +65 64789047
>> www.bii.a-star.edu.sg/~cpchng
>>
>>
>> On 6/22/04 11:50 AM, "Dr Seth Carlton OLSEN" <s.olsen1 at uq.edu.au>
>> wrote:
>>> Hi Gromacs Users,
>>> I am trying to install GROMACS on in Intel Pentium 4 machine
>> running Fedora
>>> Core 2 Linux.  When I unpack the rpm, I get the following
>> complaint regarding
>>> FFTW dependencies:
>>>
>>> error: Failed dependencies:
>>>       libsfftw.so.2 is needed by gromacs-3.2.1-1
>>>       libsrfftw.so.2 is needed by gromacs-3.2.1-1
>>>
>>> Upon checking in /usr/lib, I find the following FFTW related
>> libraries:>
>>> /usr/lib/libfftw.so.2              /usr/lib/librfftw.so.2
>>> /usr/lib/libfftw.so.2.0.7          /usr/lib/librfftw.so.2.0.7
>>> /usr/lib/libfftw_threads.so.2      /usr/lib/librfftw_threads.so.2
>>> /usr/lib/libfftw_threads.so.2.0.7
>> /usr/lib/librfftw_threads.so.2.0.7>
>>> I'm not sure what to do at this point (new Fedora user), but my
>> instinct is to
>>> make the following links in /usr/lib:
>>>
>>> /usr/lib/libfftw.so.2 --> libsfftw.so.2
>>> /usr/lib/librfftw.so.2 --> libsrfftw.so.2
>>>
>>> I expect I will try this, and post to the list whether or not it
>> works.> However, if anyone who reads this knows another, cleaner
>> way of circumventing
>>> the problem, I would appreciate the knowledge.  Thanks!
>>>
>>> Cheers,
>>> Seth
>>>
>>>
>>>
>>> ccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccms
>>>
>>> Dr Seth Olsen, PhD
>>> Postdoctoral Fellow, Computational Systems Biolgy Group
>>> Centre for Computational Molecular Science
>>> Chemistry Building,
>>> The University of Queensland
>>> Qld 4072, Brisbane, Australia
>>>
>>> tel (617) 33653732
>>> fax (617) 33654623
>>> email: s.olsen1 at uq.edu.au
>>> Web: www.ccms.uq.edu.au
>>>
>>> ccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccms
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> gmx-users mailing list
>>> gmx-users at gromacs.org
>>> http://www.gromacs.org/mailman/listinfo/gmx-users
>>> Please don't post (un)subscribe requests to the list. Use the
>>> www interface or send it to gmx-users-request at gromacs.org.
>>>
>>
>> _______________________________________________
>> gmx-users mailing list
>> gmx-users at gromacs.org
>> http://www.gromacs.org/mailman/listinfo/gmx-users
>> Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-users-request at gromacs.org.
>>
>
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.