On Fri, May 11, 2012, Cao Yang wrote:
> I found that AMBER is much slower than the official's test on my
> machine. Details are as follows.
You are using a value of cut of 10.0, which will slow down the calculation
a lot compared to the benchmarks. You could also test how the calculation
scales with number of threads: sometimes using all the threads on a node
is not optimal, since any kind of background process might cause
synchronization problems. Try running with 10 threads just to see what
happens.
....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri May 11 2012 - 05:30:04 PDT
