RE: AMBER: Error - must define AMBERHOME environment variable

From: Ross Walker <ross.rosswalker.co.uk>
Date: Tue, 27 Mar 2007 08:31:26 -0700

Dear Bertrand,
 
Firstly on an sgi_altix you should use:
 
./configure -nopar sgi_altix (for the non-parallel version, best to get
this working first)
 
and
 
./configure -mpi sgi_altix (for the parallel version.)
 
this way the correct libraries and compiler flags should be set.
 
The segfault is coming from the compiler and is a bug in the compiler itself
and there is not a lot we can do about it other than try to work around it.
This is Intel's fault and they should fix it. The problem is made more
complicated by the fact that some versions of the compiler work fine while
others crash. And even worse when you go to a newer compiler some previous
bugs are fixed and yet newer ones are introduced :-(.
 
Firstly though, what version of the compiler are you using? You should go
and get the very latest compiler just in case some of the bugs may have been
fixed.
 
Next obtain bugfix.all from the bugfixes page on amber.scripps.edu and patch
your source code with it.
 
Then cd $AMBERHOME/src
 
make clean
make
 
See how far the compiler gets before it segfaults. The following only
applies to the compilation of files in $AMBERHOME/src/sander. If the
compiler dies on a file outside this directory then please send me the make
log. If it segfaults on a specific file in src/sander then
 
cd $AMBERHOME/src/sander
 
edit that file and remove the first line of the file that reads "!
<compile=optimized>"
 
Next do "rm -f depend"
makedepend > depend
cd ..
make
 
Hopefully it will proceed beyond that file and perhaps segfault somewhere
else or maybe compile successfully. If it still segfaults on the same file
then edit $AMBERHOME/src/config.h and change
 
FFLAGS= -w95 -ftz -O2 $(LOCALFLAGS) $(AMBERBUILDFLAGS)
to
FFLAGS= -w95 -ftz -O0 $(LOCALFLAGS) $(AMBERBUILDFLAGS)
 
Then try make again.
 
Hopefully this will allow you to build a working copy of sander. Next run
the test cases and see how it does. Let me know if it fails on any,
particularly the QMMM ones since some versions of the Intel compilers had
problems with qm_link_atoms.f then let me know and I can explain how to work
around this compiler bug as well.
 
Failing all of the above see if you can get hold of ifort v8.1.025 for
Itanium (IA64) since this version I know to work out of the box.
 
Good luck....
 
Ross
 
/\
\/
|\oss Walker

| HPC Consultant and Staff Scientist |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
| http://www.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 read every day, and should not be used for urgent or sensitive issues.
 


  _____

From: owner-amber.scripps.edu [mailto:owner-amber.scripps.edu] On Behalf Of
bertrand russell
Sent: Monday, March 26, 2007 22:58
To: amber.scripps.edu
Subject: Re: AMBER: Error - must define AMBERHOME environment variable


Dear Mr.Ross,

                      Thank you for your support. I herewith sending you the
compiler error message which I got while compiling with ifort_ia64 compiler.
The steps I followed are and the error message is as follows,

>cd $AMBERHOME/src
>./configure -p4 ifort_ia64
After this point I edited the the config.h file by adding "-ldl" flag
immediately following ifort wherever it is in the config.h file. Then I
issued the following command,
>make serial
My error messages are as follows,

##########################################################
ifort -ldl -c -w95 -ftz -O2 -FR -o mdread.o _mdread.f
cpp -traditional -P rdparm.f > _rdparm.f
ifort -ldl -c -w95 -ftz -O2 -FR -o rdparm.o _rdparm.f
cpp -traditional -P getcor.f > _getcor.f
ifort -ldl -c -w95 -ftz -O2 -FR -o getcor.o _getcor.f
cpp -traditional -P sa_driver.f > _sa_driver.f
ifort -ldl -c -w95 -ftz -O3 -IPF_fma -IPF_fp_relaxed -ip -FR -o
sa_driver.o _sa_driver.f
cpp -traditional -P pb_force.f > _pb_force.f
ifort -ldl -c -w95 -ftz -O3 -IPF_fma -IPF_fp_relaxed -ip -FR -o
pb_force.o _pb_force.f
fortcom: Severe: _pb_force.f, line 1597: **Internal compiler error:
segmentation violation signal raised** Please report this error along with
the circumstances in which it occurred in a Software Problem Report. Note:
File and line given may not be explicit cause of this error.
end function ey
^
compilation aborted for _pb_force.f (code 3)
make[1]: *** [pb_force.o] Error 3
make[1]: Leaving directory `/pludisk1/oth/amber/amber9/src/pbsa'
make: *** [serial] Error 2
bash-2.05b$
########################################################

If I reduce the optimisation level to "O2" instead of "O3" I am getting the
same error but at 1616 line number instaead of 1597th line. It would be
helpful if I can solve this problem and install amber9 in our SGI Altix. I
patched with the latest bugfix.all file from amber.scripps.edu site.

My machine specifications:
Machine Name: Altix 390
Processor Type: Intel Itanium
Operating system: ALTIX SGI Propack
ifort compiler version: Version 8.0

Many thanks in advance.





-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Mar 28 2007 - 06:07:41 PDT
Custom Search