Re: amber-developers: ti_ggcc tests.. vlimit errors?

From: Thomas E. Cheatham, III <>
Date: Fri, 31 Mar 2006 00:58:29 -0700

> Have you tested this for side-effects? If not, how are you solving the
> problem at your site?

I've run all the tests with gfortran and pgf90 with my small changes to
files.h and do not see any errors related to that (but I do see others
as outlined below); (we were solving the problem currently by using
symbolic links or copying files around to directories with smaller paths).

p.s. DAC, I can get accounts for Ross and Mike if this will help with
pathscale access...


Other testing/compiling issues, snapshot 28mar:

** pentium, pgf90, -bintraj (configure -bintraj pgf90)

 -- bintraj.f cannot find typesizes.mod (as a result of use netcdf)
    solution: -I../netcdf/include added to AMBERFLAGS in config.h
    [is this the best place for this? Maybe add to FFLAGS]

    TESTS: all tests pass with -bintraj or files.h modifications

** opteron, gfortran, -bintraj (configure -bintraj gfortran)

 -- mismatch with underscores between netcdf and amber
    tried and did not work:
     (1) remove -fno-second-underscore everywhere
     (2) selectively add -fsecond-underscore
     (3) change -Df2cfortrain to -DpgiFortran

    solution: ? (idea's anyone???)

** opteron | pentium, pathscale, -bintraj

 -- configure for netcdf fails; pathf90 does not produce an executable,
    down the road you get messages like:

    make[4]: Entering directory
    pathcc -c -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -O1 -I../libsrc
-DNDEBUG -DpgiFortran fort-attio.c
    In file included from fort-attio.c:6:
    ncfortran.h:27:42: NF_INT_IS_C_... not defined: No such file or

    Problem: Compiling the test program by the netcdf ./configure
      fails due to the high optimization (foptflags) and then switches to
      any other available compiler automagically (pgf90 in our case).
      Replacing the foptflags by fflags fixes it.

    Problem 2: No setting of NETCDF.mod

    Problem 3: F/C linking, second underscore causes link failure

    Problem 4: ptraj/rdparm: Need to define dependency for NETCDF.mod to
    match netcdf.mod

    SOLUTION: Modify f90ncflags/fncflags to be "fflags" rather than
      in configure, netcdf section

    SOLUTION2: Add if bintraj netcdf=NETCDF.mod check in pathscale section

    SOLUTION3: Add -fno-second-underscore to foptflags, fflags, cflags in
    pathscale section of configure.

    SOLUTION4: Add NETCDF.mod:: dependency like netcdf.mod

** opteron | pentium, pathscale (-bintraj or no -bintraj)

 -- lot's of warnings in ameoba compiling:

pathf90-1542 pathf90: WARNING XFORM_MPOLE_MATRIX, File =
_amoeba_multipoles.f, Line = 851, Column = 8
  Initializing a named common block member in a program unit other than a
block data is not legal for this architecture.

  data hind4 /1,2,3,2,3,1,3,1,2,2,3,3,3,3,2/

** opteron, pathscale

 -- internal compiler error on force.PIMD
    dropping optimization to -O2 allows it to continue.

** opteron, gfortran (no -bintraj)

 -- compiles fine, but test error with ./Run.noesy
    [I do not know how to fix this]

    From noesy.out:

          Amber 9 SANDER 2006

| Run on 03/30/2006 at 12:26:55
  [-O]verwriting output

File Assignments:
| MDIN: mdin
| MDOUT: noesy.out
| PARM: prmtop
|RESTRT: restrt
| REFC: refc
| MDVEL: mdvel
| MDEN: mden
| MDCRD: mdcrd
|MDINFO: mdinfo
|INPDIP: inpdip
|RSTDIP: rstdip

 Here is the input file:

 just get final values
    ntx=1, ntb=0,
    pencut= 0.1, nmropt=2, ipnlty=1,
    imin=1, maxcyc=1,
    ntc=1, ntf=1, cut=9.0,
    mxsub=10, tausw=0.101,

 &wt type='NOESY', istep1=0,istep2=500,value1=1.0,
            value2=1.0, /

 &wt type='END' /


           Begin reading energy term weight changes/NMR restraints
 NOESY 0 500 1.000000 1.000000 0 0

 Requested file redirections:
  NOESY = NXP.8May95
  DISANG = RST.16sept93
 Restraints will be read from file: RST.16sept93
Here are comments from the DISANG input file:
# 1 GUA h1p 1 GUA h2d 2.7 (#141

                       Number of restraints read = 590

                  Done reading weight changes/NMR restraints

 Noesy volumes will be read from file: NXP.8May95
 Here are comments from the NOEsy input file:
 # first pass at fixed-distance file

 Namelist reports error in reading noeexp
 -- Subscript out of range implies dimensioning problem
 -- (see nmr.h)
Received on Wed Apr 05 2006 - 23:49:33 PDT
Custom Search