Re: AMBER: Compiling using pgf90

From: Craig TOEPFER <craig.toepfer.st.com>
Date: Wed, 06 Oct 2004 09:52:06 -0700

David,

You need to set your PATH to:

/usr/pgi/linux86-64/5.2/bin

such that when you issue the command "which pgf90" you should see:

/usr/pgi/linux86-64/5.2/bin/pgf90

I also recommend for best performance that the "-fastsse" switch be used
for compilation. This can be done when you invoke make as follows:

make LOCALFLAGS="-fastsse" serial

-craig


David Lomelin wrote:
> Amber users:
>
> I have been able to compile and successfully run Amber8 using
> Intel?s ifort compiler and Math Kernel Library; however, I was told that
> the Portland Group?s compilers might improve performance, so I have been
> trying to recompile Amber8 using the Portland Group compilers. I have a
> dual processor opteron system using Fedora Core 2 ? I have installed
> MPICH but I have not tried compiling Amber8 using those libraries
> though. After using ?configure ?opteron pgf90? and running ?make
> serial?, it successfully compiles but when I try and run a simple
> minimization, I get a segmentation fault error. I found that the
> configure script had the directory of the ACML and LAPACK libraries
> hardcoded and did not point at the correct directory, so I changed it to
> my local installation:
>
> if [ -e $PGI/linux86/5.2/lib/liblapack.a ]; then ?
>
> instead of:
>
> if [ -e /usr/pgi/lib/liblapack.a ?.
>
> This did not fix the problem however.
>
> The Portland group?s website offers alternate versions of the configure
> script that should fix any problems, but I have not been able to get
> them to work. http://www.pgroup.com/resources/amber/amber8_pgi52.htm
> When I use their configure script and run ?configure ?nopar ?opteron
> ?acml ?lapack pgf90? followed by ?make serial?, I get the following
> error, after which compilation halts:
>
> .
>
> .
>
> .
>
> .
>
> pgf90 -Mfree -o sander trace.o lmod.o decomp.o icosasurf.o egb.o
> findmask.o pb_force.o sander.o cshf.o noecalc.o noeread.o caldis.o
> calrate.o dinten.o drates.o indexn.o kmat.o pearsn.o plane.o remarc.o
> nmrcal.o nmrred.o restal.o getnat.o nmrnrg.o modwt.o disnrg.o angnrg.o
> tornrg.o nmrprt.o nmrgrp.o nmrcms.o nmrcmf.o impnum.o nmrsht.o at2res.o
> chklin.o opnmrg.o printe.o runmin.o ndvprt.o force.o rdparm.o mdread.o
> locmem.o runmd.o getcor.o r6ave.o r6drv.o aveint.o degcnt.o corf.o
> threeb.o tripl.o nmrrad.o decnvh.o fastwt.o echoin.o parallel.o jnrg.o
> shake.o ene.o mdwrit.o minrit.o set.o setmm.o dynlib.o mdfil.o nmlsrc.o
> ew_force.o ew_setup.o ew_box.o ew_bspline.o ew_fft.o ew_direct.o
> ew_recip.o pcshift.o align.o rstack.o istack.o rfree.o rgroup.o random.o
> lsqfit.o amopen.o debug.o ew_recip_reg.o ew_handle_dips.o
> ew_dipole_recip.o mexit.o new_time.o extra_pts.o thermo_int.o matinv.o
> assert.o mmtsb.o mmtsb_client.o erfcfun.o veclib.o mdm.o pb_init.o
> constantph.o prn_dipoles.o \
>
> ../lmod/lmod.a -lacml -llapack -lblas \
>
> ../lapack/lapack.a ../blas/blas.a ../lib/nxtsec.o ../lib/sys.a
>
> icosasurf.o(.text+0x1cb4): In function `icosasurf_gen_rot_mat_':
>
> : undefined reference to `__mth_i_dcosx'
>
> icosasurf.o(.text+0x1ccb): In function `icosasurf_gen_rot_mat_':
>
> : undefined reference to `__mth_i_dsinx'
>
> egb.o(.text+0x1ba3): In function `genborn_egb_':
>
> : undefined reference to `__mth_i_dlogx'
>
> egb.o(.text+0x1ef1): In function `genborn_egb_':
>
> : undefined reference to `__mth_i_dlogx'
>
> .
>
> .
>
> .
>
> .
>
> Any help would be appreciated. Thank you.
>
> --David
>


-- 
------------------------------------------------------------------------
Craig Toepfer                                    craig.toepfer.st.com
Advanced Compilers and Tools                     http://www.wil.st.com
AST Portland Lab, STMicroelectronics             Phone:  +1-503-682-2806
9150 SW Pioneer Ct, Suite H                      FAX:    +1-503-682-2637
Wilsonville, OR USA 97070
------------------------------------------------------------------------
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Oct 06 2004 - 18:53:00 PDT
Custom Search