Re: AMBER: Leap compilation issues

From: justin litchfield <litch.stanford.edu>
Date: Mon, 11 Jul 2005 18:25:10 -0700

On Jul 11, 2005, at 10:05 AM, Ross Walker wrote:

> Is the path /opt/intel/ifc/lib the correct compiler version? The
> reason I
> ask is that Intel switched to the name ifort for version 8 and
> later. Hence
> I suspect you may be setting LD_LIBRARY_PATH to a different
> compiler version
> to the one you compiled with.
yes, that's the right compiler version, i checked it with which,
etc. i had named it something non-standard on the install.

> There are two options:
>
> 1. Re-run amber's configure script with the -static option and then
> recompile.
this fails on recompile, with the most applicable error message
similar to this one appearing several times:
/opt/intel/ifc/lib/libifcore.a(for_open_proc.o)(.text+0x3826): In
function `for__compute_filename':
../src/libfor/for_open_proc.c: warning: Using 'getpwnam' in statically
linked applications requires at runtime the shared libraries from the
glibc version used for linking

complete log in serial_p4_ifort.txt


> 2. Make sure you source /opt/intel_fc_80/bin/ifortvars.sh -> Or
> the correct
> path for your compiler version. BEFORE you compile and BEFORE you run
> Amber... Alternatively add this to your .bashrc file.
sourcing this before compiling then running sander gives the same
error. make test.leap works (until it gets to the sander parts), and
xleap works, but antechamber (divcon) fails, as does sander.
attached are the antechamber/sander failures. also, note that in the
compilation there are tons of warnings like this:
"xaAtomTable.c:636: warning: cast to pointer from integer of
different size"

a third attachment has my compilers/versions. andreas, you had
suggested using maybe a different compiler or something with FC4,
does anyone have specific suggestions for what to use/how to do
this? i saw that forcing it to compile in 32 bit mode should be
possible, but does anyone have specific directions? thank you,

justin




>
> All the best
> Ross
>
> /\
> \/
> |\oss Walker
>
> | Department of Molecular Biology TPC15 |
> | The Scripps Research Institute |
> | Tel: +1 858 784 8889 | EMail:- ross.rosswalker.co.uk |
> | http://www.rosswalker.co.uk | PGP Key available on request |
>
> Note: Electronic Mail is not secure, has no guarantee of delivery,
> may not
> be ready every day, and should not be used for urgent or sensitive
> issues.
>
>
>> -----Original Message-----
>> From: owner-amber.scripps.edu
>> [mailto:owner-amber.scripps.edu] On Behalf Of justin litchfield
>> Sent: Sunday, July 10, 2005 23:50
>> To: amber.scripps.edu
>> Subject: Re: AMBER: Leap compilation issues
>>
>> i am definitely not very informed on these matters, but was able to
>> get leap working. i just had to install one more X11-dev
>> library. i
>> did get everything compiled using ifc8.1. i got ifc 9.0 and found a
>> thread that talks about how one can manually install the 64-bit
>> compilers, but i don't have the expertise to get that to work that
>> really. i'm a bit over my head here. i'm not really sure about
>> which parts were 64 bit and which are 32 bit. xleap and tleap both
>> seem to work, but sander seems to be broken. the first error was
>>
>> [root.jd10 test]# make test
>> cd dmp; ./Run.dmp
>> ../../exe/sander: error while loading shared libraries: libsvml.so:
>> cannot open shared object file: No such file or directory
>> ./Run.dmp: Program error
>> make: *** [test.sander.no_lmod] Error 1
>>
>> so i googled that problem and found that this could be fixed
>> by doing
>> this:
>> [root.jd10 test]# export LD_LIBRARY_PATH=/opt/intel/ifc/lib
>> at which point i get this
>>
>> [root.jd10 test]# make test
>> cd dmp; ./Run.dmp
>> forrtl: severe (174): SIGSEGV, segmentation fault occurred
>> Image PC Routine Line Source
>> sander 0814EE20 Unknown Unknown Unknown
>> sander 08077776 Unknown Unknown Unknown
>> sander 0804A398 Unknown Unknown Unknown
>> Unknown 00449DE6 Unknown Unknown Unknown
>> sander 0804A251 Unknown Unknown Unknown
>> ./Run.dmp: Program error
>> make: *** [test.sander.no_lmod] Error 1
>> [root.jd10 test]#
>>
>> sorry if this is jumbled or confusing. at least this is excititg,
>> and i got leap working. i'll keep working on this. thank you all,
>>
>> justin
>>
>>
>> On Jul 10, 2005, at 10:13 PM, David A. Case wrote:
>>
>>
>>> On Sun, Jul 10, 2005, justin litchfield wrote:
>>>
>>>
>>>
>>>> i am now trying to install amber on a new Athlon 64 running FC4.
>>>> everything seems to go great, i'm using gcc 4, ifc 8.1, but when it
>>>> gets to the leap installation it crashes out.
>>>>
>>>>
>>>
>>>
>>>
>>>> MapAg.c:190: warning: cast from pointer to integer of
>>>>
>> different size
>>
>>>> MapAg.c:209: warning: cast from pointer to integer of
>>>>
>> different size
>>
>>>>
>>>>
>>>
>>>
>>>
>>>> WcActCB.c:1536: error: syntax error before ???}??? token
>>>> WcActCB.c:1538: error: syntax error before ???(??? token
>>>>
>>>>
>>>
>>> I know this is question that mainly reflects by ignorance, but is
>>> "Athlon 64
>>> running FC4" a 64-bit OS (where pointers are 64-bit). xleap is
>>> probably
>>> broken on such a platform. See the comments in $AMBERHOME/src/
>>> Makefile
>>> on how to compile for tleap only (which seems to be 64-bit clean).
>>>
>>> If any amber users know the answers to the following questions,
>>> please post
>>> them:
>>>
>>> 1. Has anyone gotten xleap to run in 64-bit mode? How?
>>>
>>> 2. Are there compile/load options that can force a 32-bit build?
>>>
>>> 3. If you take an xLeap executable from an IA-32 machine, and try
>>> to run it
>>> on a 64-bit machine, will it work?
>>>
>>> 4. Has anyone used gcc 4 to compiler xaLeap on a 32-bit system?
>>>
>>> 5. Has anyone every seriously looked a compiler warnings
>>>
>> like those
>>
>>> cited
>>> above, and thought about whether things could be fixed?
>>>
>>> ...thanks!....dac
>>>
>>>
>> --------------------------------------------------------------
>> --------
>>
>>> -
>>> The AMBER Mail Reflector
>>> To post, send mail to amber.scripps.edu
>>> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>>>
>>>
>>
>> --------------------------------------------------------------
>> ---------
>> The AMBER Mail Reflector
>> To post, send mail to amber.scripps.edu
>> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>>
>
>
> ----------------------------------------------------------------------
> -
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>


-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu

Received on Tue Jul 12 2005 - 02:53:01 PDT
Custom Search