charmm scalability on 2.4 kernels

Peter H. Koenig Peter.Koenig at
Wed Jan 16 02:36:50 PST 2002

 Bogdan Costescu wrote:
 > That is actually what I have observed during the last 3 years of running
 > different versions of kernels, MPI libraries and CHARMM. Running using
 > only one transport (TCP or shared mem) is always better than mixing them,
 > f.e (using LAM-6.5.6):
 > CPUs    nodes   real time (min)         transports
 > 4       4       5.95                    TCP
 > 4       2       7.08                    TCP+USYSV
 > As you can see, the difference is quite significant.
 Do you also have the numbers for 2 dual nodes using only TCP ?
 Peter Koenig

More information about the Beowulf mailing list