AMBER compiles with reasonable success, except for a ton of warnings
about problems between 32-bit and 64-bit pointer casts, which arise
because the gcc compiler I am using doesn't support the -taso option.
Will this cause me any problems later? Has anyone found a workaround for
this?
However, when I test the compilation by typing 'make test.all' in the
$AMBERHOME/test directory, the program chokes and dies in the Run.nmr
routine, during a call to sander, because of a floating point error. The
failure seems to take place in the FORTRAN code for setting machine
precision, etc. The debugger reports the following:
object file name: ../../exe/sander
Reading symbolic information ...done
(ladebug) run -O -i mdin.nmr -o mdout
Thread received signal FPE
stopped at [subroutine dlamc1():221 0x1200aea48]
Source file not found or not readable, tried...
dlamch.f
./dlamch.f
../../exe/dlamch.f
(Cannot find source file dlamch.f)
(ladebug) where
>0 0x1200aea48 in dlamc1() "dlamch.f":221
#1 0x1200aeabc in dlamc2() "dlamch.f":424
#2 0x1200ae66c in dlamch() "dlamch.f":76
#3 0x1200a959c in dspev() "dspev.f":132
#4 0x12008ba7c in align1() "_align_.f":377
#5 0x12002a61c in force() "_force_.f":811
#6 0x120039c0c in runmd() "_runmd_.f":637
#7 0x12000356c in sander() "_sander_.f":822
#8 0x1200a9314 in ../../exe/sander
#9 0x2000156bd40
#10 0x11ffff70c
Does anyone have any idea what is wrong here? Another floating point
error happens during the call to nmode, if that sheds any light on the
problem. I have applied all the bug fixes that might bear on the problem
(at least I think so). If there is more info I can supply, please
let me know. Thanks in advance for the help!
Elizabeth White
Received on Thu May 31 2001 - 14:48:12 PDT
