Hi Jason, Thanks for the note (and for writing that page - I was able to find
it even before you posted) but while it describes some of the errata I found
previously, it doesn't help me, unless the Intel Fortran compiler is less
discriminating than the gnu one. I assumed that an older gcc/g++ (I used
4.7.2) would be more compatible than a more recent one 4.8.1, but maybe I
didn't go back far enough..?
OMG!! Dropping back to gcc 4.4.7 DOES WORK (??!??!), at least for the serial
version. I'll have to build a new version of openmpi for this version of gcc,
but at least there's movement forward.
Thanks!
hjm
On Tuesday, September 24, 2013 07:31:39 PM Jason Swails wrote:
> I was able to do it following these instructions:
> http://jswails.wikidot.com/installing-amber11-and-ambertools-1-5
>
> Can you try those?
>
> HTH,
> Jason
>
> On Tue, Sep 24, 2013 at 3:19 PM, harry mangalam
<harry.mangalam.uci.edu>wrote:
> > Hi All,
> >
> > I'm trying to compile amber11 for a faculty member.
> >
> > The environment is 64b CentOS 6.4 with the following environment modules
> > loaded:
> >
> > Currently Loaded Modulefiles:
> > 1) gmp/5.0.5 3) mpfr/3.1.1 5) gcc/4.7.2
> > 2) mpc/1.0.1 4) binutils/2.23.1 6) netcdf/4.2.1.1-share-gcc
> >
> > My AMBERHOME is set to /data/apps/amber/amber11
> >
> > I've been given the amber11 and AmberTools tarballs and have patched them
> > according to this page: <http://ambermd.org/bugfixes11.html>
> >
> > This patching succeeds on most patches; but some fail:
> >
> > Tue Sep 24 12:08:46 [0.24 0.27 0.59] root.hpc-s:/data/apps/amber/amber11
> > 1263 $ find . -name \*.rej
> > ./src/pmemd/src/cuda/kCCF.h.rej
> > ./src/pmemd/src/cuda/gpu.cpp.rej
> > ./src/pmemd/src/cuda/gputypes.h.rej
> > ./src/pmemd/src/cuda/kBNL.h.rej
> > ./src/pmemd/src/cuda/cuda_info.fpp.rej
> > ./src/pmemd/src/cuda/gputypes.cpp.rej
> > ./src/pmemd/src/cuda/kShake.cu.rej
> > ./src/pmemd/src/cuda/kNeighborList.cu.rej
> > ./src/pmemd/src/pairs_calc_vec.i.rej
> > ./src/pmemd/src/master_setup.fpp.rej
> > ./src/pmemd/src/runfiles.fpp.rej
> > ./src/pmemd/src/extra_pnts_nb14.fpp.rej
> > ./src/pmemd/src/gb_ene.fpp.rej
> > ./src/pmemd/src/pairs_calc_novec.i.rej
> > ./src/sander/egb.f.rej
> > ./src/sander/extra_pts.f.rej
> > ./src/sander/nonbond_list.f.rej
> > ./src/sander/short_ene.f.rej
> > ./AmberTools/src/configure.rej
> >
> > None of these is in the $AMBERHOME/AmberTools/src/pbsa dir
> >
> > I've tried to compile the AmberTools as described and with a little
> > munging of
> > paths, replacing of archaic libtools, and interjecting of other libs (ie.
> > our
> > local netcdf module) have managed to compile it to ~ the same point
> > described
> > here: <http://archive.ambermd.org/201111/0442.html>
> >
> > My compilation ends like:
> > Tue Sep 24 12:07:19 [0.18 0.27 0.62] root.hpc-
> > s:/data/apps/amber/amber11/AmberTools/src/pbsa
> > 1261 $ make
> > cpp -traditional -P -DBINTRAJ amg1r5.f > _amg1r5.f
> > gfortran -c -O3 -mtune=generic -ffree-form -o amg1r5.o _amg1r5.f
> >
> > _amg1r5.f:1203.16:
> > ifg,a(jtrst),time,ndjtr,ium,mdjtr)
> >
> > 1
> >
> > Warning: Type mismatch in argument 'jtr' at (1); passed REAL(8) to
> > INTEGER(4)
> >
> > _amg1r5.f:1216.58:
> > call pcol(k-1,ierr,ia,ja,iw,imin,imax,imaxw,icg,ifg,u(irst), &
> >
> > 1
> >
> > Warning: Type mismatch in argument 'ir' at (1); passed REAL(8) to
> > INTEGER(4)
> >
> > _amg1r5.f:1223.26:
> > imaxw,ifg,icg,u(irst),time,ierr,irow0,ncolx, &
> >
> > 1
> >
> > Warning: Type mismatch in argument 'ncolor' at (1); passed REAL(8) to
> > INTEGER(4)
> >
> > _amg1r5.f:1231.16:
> > ifg,u(irst),nstcol,ncolx,time,nda,ndja,ium, &
> >
> > 1
> >
> > Warning: Type mismatch in argument 'ncolor' at (1); passed REAL(8) to
> > INTEGER(4)
> >
> > _amg1r5.f:3460.23:
> > external idec,random
> >
> > 1
> >
> > Error: Return type mismatch of function 'random' at (1) (REAL(8)/REAL(4))
> >
> > _amg1r5.f:3028.45:
> > ja(jlo),a(jlo),f(ilo),u(ilo),nsp,a(jhi+1),a(jhi+1), &
> >
> > 1
> >
> > Warning: Type mismatch in argument 'isp' at (1); passed REAL(8) to
> > INTEGER(4)
> >
> > _amg1r5.f:3094.45:
> > ja(jlo),a(jlo),f(ilo),u(ilo),nsp,a(jhi+1),a(jhi+1), &
> >
> > 1
> >
> > Warning: Type mismatch in argument 'isp' at (1); passed REAL(8) to
> > INTEGER(4)
> >
> > _amg1r5.f:3618.17:
> > external cgalf,cgeps,ctime
> >
> > 1
> >
> > Error: Return type mismatch of function 'cgalf' at (1) (REAL(8)/REAL(4))
> >
> > _amg1r5.f:3618.23:
> > external cgalf,cgeps,ctime
> >
> > 1
> >
> > Error: Return type mismatch of function 'cgeps' at (1) (REAL(8)/REAL(4))
> > make: *** [amg1r5.o] Error 1
> >
> > The followup AMBER list posting suggests applying the patches to the
> > source
> > tree, but I've already done this tho without perfect success.
> >
> > So, is there an easy fix that I can apply via edit, or is there a
> > completely
> > patched source code tree that I can D/L to try compile. Also, we have
> > most of
> > the AmberTools libs already installed; is there an easy way to ignore the
> > AmberTools tree and use our own libs?
> >
> > Also, I see that amber12 is also available. The faculty is licensed for
> > amber11, but does amber12 require another fee?
> >
> > Thanks in advance
> > Harry
> >
> > ---
> > Harry Mangalam - Research Computing, OIT, Rm 225 MSTB, UC Irvine
> > [m/c 2225] / 92697 Google Voice Multiplexer: (949) 478-4487
> > 415 South Circle View Dr, Irvine, CA, 92697 [shipping]
> > MSTB Lat/Long: (33.642025,-117.844414) (paste into Google Maps)
> > ---
> >
> >
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
---
Harry Mangalam - Research Computing, OIT, Rm 225 MSTB, UC Irvine
[m/c 2225] / 92697 Google Voice Multiplexer: (949) 478-4487
415 South Circle View Dr, Irvine, CA, 92697 [shipping]
MSTB Lat/Long: (33.642025,-117.844414) (paste into Google Maps)
---
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Sep 24 2013 - 19:30:03 PDT
