Dear list members,
Now I am begining a QM/MM study by using the
ROAR program. However, in testing the performance
of the program by calculating the sample demo of HCAII at
demo/roar/coupled/ , I met the folloing trouble.
"escf" values calculated on xp1000 workstations
largely differ from the correct values (in demo_qmmm.out),
whereas values calculated on SGI machines show only
little(?) difference.
For example, the first 10 esf values for both machines are ...
----- SGI -----
escf --> -188.6457639171858
escf --> -188.8235439910495
escf --> -189.1166301830090
escf --> -189.5212588869908
escf --> -190.0419032338541
escf --> -190.5917845003423
escf --> -191.1881699582736
escf --> -191.8229450605722
escf --> -192.5340733262710
escf --> -193.2396915067802
----- XP1000
escf --> -188.655855084537
escf --> -187.775560038339
escf --> -183.622990995282
escf --> -174.191115104564
escf --> -158.593207019614
escf --> -136.536892595410
escf --> -107.956239101855
escf --> -73.2455634161743
escf --> -33.9538729554479
escf --> 7.49879035851336
and values in the sample output file are...
----- demo_qmmm.out -----
escf --> -188.1682744101126
escf --> -188.3459198068158
escf --> -188.6388413305249
escf --> -189.0433030630520
escf --> -189.5638616754150
escf --> -190.1137508591491
escf --> -190.7101882513962
escf --> -191.3450183429086
escf --> -192.0563241936179
escf --> -192.7619471662765
Would anyone please tell me the proper way of
succeeding in calculation or compilation on axp
machines? I tried both f77 and f90 types of
compilation for reference.
Thanks in advance.
--
Hajime HIRAO
Nagoya University
Received on Tue May 30 2000 - 19:23:02 PDT
