G03-B05/linda parallel for large system and multiple-processors

Dear CCLers,

I am having some troubles in using many processors, %NProcLinda= 8 or more,
for large molecular systems (more than than 60 atoms) for G03/linda. In those
cases, only one node work and other nodes assigned to are idle all the time.
But for small jobs(less atoms) with less CPU procesors, like %NProcLinda=4 or
less, the job is paralleling well. Does anybody have any experience about this
G03/linda parallel problem? Our computer system are PC clusters with 128
Dual-Xeon Nodes. The details are shown in the following pages. We are using
PBS for job sumitting.

http://www.hku.hk/cc_news/ccnews106/hpcpower.htm
http://www.hku.hk/cc/ccsystem/hpcpower

Thanks in advance.

P.S. Our administrator can not solve the problem right now and wish to get a
solution from the vendor.

Dr. Cunyuan Zhao
Department of Chemistry
The University of Hong Kong, Hong Kong
Email: zhaocy!at!hkucc.hku.hk
Tel:(852)2859 8947