AMBER Archive (2006)

Subject: AMBER: Amber 8 QMMM tests fail on AIX

From: Shan-ho Tsai (tsai_at_hal.physast.uga.edu)
Date: Thu Jun 01 2006 - 10:08:26 CDT

Hello all,

I have compiled Amber8 on an IBM power4 cluster running
AIX 5.2, using XLF 8.1.1 and the following steps:

./configure xlf90_aix
make

(applied bugfix.all that includes 61 bug fixes).
I didn't get any errors in the compilation, but the qmmm
tests seem to fail drastically. "make test" ends with

cd qmmm/standard; ./Run.lysine
STOP 0
diffing lysine.out.save with lysine.out
possible FAILURE: check lysine.out.dif
==============================================================
cd qmmm/divcon; ./Run.crambin
Segmentation fault
  ./Run.crambin: Program error

(All other tests passed.)

Here are some results that I obtained in lysine.out:

   NSTEP ENERGY RMS GMAX NAME NUMBER
      1 -1.3941E+24 4.4401E+01 1.8941E+02 HG2 9

 BOND = 0.0016 ANGLE = 0.3736 DIHED = 0.0026
 VDWAALS = 0.1908 EEL = -5.8408 HBOND = 0.0000
 1-4 VDW = 3.7051 1-4 EEL = 61.7356 RESTRAINT = 0.0000
 ESCF = *************

And below are the corresponding lines in lysine.out.save:

   NSTEP ENERGY RMS GMAX NAME NUMBER
      1 1.4574E+00 4.4949E+00 1.7001E+01 N 1

 BOND = 0.0016 ANGLE = 0.3736 DIHED = 0.0026
 VDWAALS = 0.1908 EEL = -5.8408 HBOND = 0.0000
 1-4 VDW = 3.7051 1-4 EEL = 61.7356 RESTRAINT = 0.0000
 ESCF = -58.7112

I get these errors when using a low optimization level (-O) and
when using -O3 -qstrict -qarch=auto -qtune=auto.

Any suggestions of how to fix these test failures will be very much
appreciated. Sorry if this has been discussed in the list before (I
haven't been able to find it).

Thank you very much,
Shan-Ho

-----------------------------------------------------------
Shan-Ho Tsai
University of Georgia, Athens, GA

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu