AMBER Archive (2009)

Subject: Re: [AMBER] pmemd on BlueGene/P?

From: Robert Duke (rduke_at_email.unc.edu)
Date: Thu Feb 05 2009 - 07:42:32 CST


Hannes -
I would expect since it works on /L, and as far as I know the architecture
is no more restrictive on /P, there should not be a problem. But I don't
have access to a BG/P, and working out the config has thus not been an
option. Perhaps someone else in the amber community with access to one of
these beasts will be able to do it. With the error you have here, it looks
like you are not even getting to compilation, but having problems with cpp
preprocessing. I think there may be multiple versions of cpp on ibm
machines (vague memory), and ibm typically has something that requires
special consideration (a nice way of saying does not work as expected by the
rest of the world). So it could be that just getting to the right cpp will
get you going. I would think carlos sosa of ibm will have a clue
(cpsosa_at_us.ibm.com).
Best Regards - Bob Duke
----- Original Message -----
From: "Hannes Loeffler" <hannes.loeffler_at_stfc.ac.uk>
To: "AMBER Mailing List" <amber_at_ambermd.org>
Sent: Thursday, February 05, 2009 8:09 AM
Subject: [AMBER] pmemd on BlueGene/P?

> Hello,
>
> does pmemd compile (and run) on a BlueGene/P? I tried to derive a
> config file from http://ambermd.org/config.bgl.html (works fine for our
> BG/L by the way) but was unsucessful in compiling.
>
> I would like to be more specific but unfortunately our compiler
> installation seems to be messed up because the cat file that would
> probably tell me what is wrong is obviously installed in the wrong
> place.
>
> /opt/ibmcmp/xlf/bg/11.1/exe/cpp -I/home/hhl/usr/include
> -I/bgsys/drivers/ppcfloor/include -DFFTW_FFT -DBINTRAJ -DMPI
> -DDIRFRC_EFS -DIRFRC_COMTRANS -DSLOW_INDIRECTVEC -DNO_C_UNDERSCORE
> parallel_dat.fpp parallel_dat.f90
> (E) Message system initialization, unable to open catalogs:
> vacfe.cat, /opt/ibmcmp/xlf/11.1/msg/en_US/vacfe.cat.
> 1506-005: (E) Error in message set 12, unable to retrieve message 296.
> make[1]: *** [parallel_dat.o] Error 1
>
>
>
> ================== first trial at config.bgp ======================
> MATH_DEFINES =
> MATH_LIBS = -L$MASSVP_LIBDIR -lmassv
> DIRFRC_DEFINES = -DDIRFRC_EFS -DIRFRC_COMTRANS -DSLOW_INDIRECTVEC
> #F90 = bgf90
> #CC = bgxlc
> #LOAD = bgf90
> F90 = mpixlf90
> CC = mpixlc
> LOAD = mpixlf90
>
> FFT_INCLUDEDIR = /home/hhl/usr/include
> FFT_LIBDIR = /home/hhl/usr/lib
> FFT_DEFINES = -DFFTW_FFT
> FFT_INCLUDE = -I$(FFT_INCLUDEDIR)
> FFT_LIBS = -L$(FFT_LIBDIR) -lfftw3
>
> NETCDF_HOME = /home/hhl/usr
> NETCDF_DEFINES = -DBINTRAJ
> NETCDF_MOD = netcdf.mod
> NETCDF_LIBS = $(NETCDF_HOME)/lib/libnetcdf.a
>
> MPI_DEFINES = -DMPI
> MPI_INCLUDE = -I/bgsys/drivers/ppcfloor/include
> MPI_LIBS = -L/bgsys/drivers/ppcfloor/lib -lmpich.rts -lmsglayer.rts
> -lrts.rts -ldevices.rts
>
> CPP = /opt/ibmcmp/xlf/bg/11.1/exe/cpp
> CPPFLAGS =
> F90_DEFINES = -DNO_C_UNDERSCORE
>
> MODULE_SUFFIX = mod
> CPUFLAGS = -qarch=450d -qtune=450
> F90FLAGS = $(CPUFLAGS) -qsuffix=f=f90 -qdpc -c
> F90_OPT_DBG = -g -O0
> F90_OPT_LO = -qmaxmem=-1
> F90_OPT_MED = -qmaxmem=-1 -O
> F90_OPT_HI = -O3 -qstrict -qhot
> F90_OPT_DFLT = $(F90_OPT_HI)
>
> CFLAGS = $(CPUFLAGS) -DNO_C_UNDERSCORE
>
> LOADFLAGS =
> LOADLIBS = -lm
>
>
> _______________________________________________
> AMBER mailing list
> AMBER_at_ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber