charmm scalability on 2.4 kernels
Peter H. Koenig
Peter.Koenig at phys.upb.de
Wed Jan 16 02:47:45 PST 2002
Bogdan Costescu wrote:
> That is actually what I have observed during the last 3 years of
> different versions of kernels, MPI libraries and CHARMM. Running
> only one transport (TCP or shared mem) is always better than mixing
> 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 ?
More information about the Beowulf