Dear Amber users;
I compiled the parallel (REM) version of amber using MPICH (1.2.5.2) on an
AMD Opteron machine. I used the ifort fortran 90 compiler and the
standard gcc compiler. I did not use the MKL libraries.
Everything compiles and it even runs!! but when I perform the tests
(test.sander.REM), it fails. There is a fairly significant difference in
the potential energy and other terms, for example:
mdout.rem.000.save:
< NSTEP = 1 TIME(PS) = 0.001 TEMP(K) = 268.60 PRESS = 0.0
< Etot = -362.5645 EKtot = 189.4862 EPtot = -552.0508
< BOND = 53.6746 ANGLE = 114.0833 DIHED = 149.5208
< 1-4 NB = 56.1429 1-4 EEL = 987.1356 VDWAALS = -114.1859
< EELEC = -1465.0603 EGB = -333.3618 RESTRAINT = 0.0000
<
------------------------------------------------------------------------------
mdout.rem.000 (my compiled executable)
> NSTEP = 1 TIME(PS) = 45646.501 TEMP(K) = 277.09 PRESS = 0.0
> Etot = -352.5541 EKtot = 195.4724 EPtot = -548.0265
> BOND = 56.6781 ANGLE = 112.3060 DIHED = 151.5083
> 1-4 NB = 57.2824 1-4 EEL = 994.1285 VDWAALS = -107.2693
> EELEC = -1491.1812 EGB = -321.4794 RESTRAINT = 0.0000
>
------------------------------------------------------------------------------
of course, the rest of the file is different as well. Any suggestions?
Thanks
-Sergio
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Tue May 17 2005 - 23:53:00 PDT
