AMBER Archive (2000)

Subject: Sander/MPI/redimension

From: Mihaly Mezei (mezei_at_inka.mssm.edu)
Date: Fri Nov 03 2000 - 12:28:27 CST

Greetings,

recently we recompiled Sander (Amber 6) with the following dimensions
(original values were left in the file commented out):

      parameter (MAXREA=2500000)
c parameter (MAXINT=2000000)
      parameter (MAXINT=2500000)
      parameter (MAXHOL=600000)
      parameter (MAXPR=5000000)
      parameter (MAXDUP=8000)
c parameter (MAX_RSTACK=1600000)
      parameter (MAX_RSTACK=6000000)
c parameter (MAX_ISTACK=100000)
      parameter (MAX_ISTACK=200000)
      parameter (MAX_STACK_PTRS=100)
      parameter (MAX_HEAP_PTRS=100)

However, when we started to run (under MPI) with 4 CPU's, within 10 minutes
we got the error message:

 ---------------------------------------------------
     Total number of mask terms = 40619
     Total number of mask terms = 81238
| Total Ewald setup time = 0.27958202
 ------------------------------------------------------------------------------

 EWALD BOMB in subroutine ew_recip/grid_weights
 nmine exceeds num_ks
 System must be very inhomogeneous. Cannot continue.

Reducing the number of CPU's to two resulted in a normal (looking) run,
wirhout any error message.

Should we have changed something eles in sizes.h?

Thanks.

Mihaly Mezei,

Dept. of Physiology & Biophysics, Mount Sinai School of Medicine
           Voice: (212) 241-2186 Fax: (212) 860-3369|
     WWW: http://www.mssm.edu/physbio/local/primfac/mezei.html