Hi all,
I found that AMBER is much slower than the official's test on my machine. Details are as follows.
SOFTWARE: AMBER11
CPU: 12*(Intel(R) Xeon(R) CPU X5690 . 3.47GHz) ; 1 node
MD: protein complex of 20872 atoms, explicit solvent. The following is the mdin file:
production2 4ns 3BWY_neg_que2_LYN_SAM_production
&cntrl
imin=0,irest=1,ntx=5,
nstlim=2000000,dt=0.002,
ntc=2,ntf=2,
cut=10.0, ntb=2, ntp=1, taup=1.0,
ntpr=5000, ntwx=5000,
ntt=3, gamma_ln=2.0,
temp0=300.0,
/
EXECUTE COMMAND: mpirun -np 12 pmemd.MPI -O -i md.in -o md.out -p md.prmtop -c eq.rst -r prod.rst -x prod.mdcrd
It takes about 1 day to finish the job. So the speed is about 4ns/day. That is much smaller than the data reported in AMBER's official website.
http://ambermd.org/gpus/benchmarks.htm
They ran the following input file with AMBER11 on 12*(Intel X5670 2.93 GHz) CPUs and the speed is 12.07ns/day.
DHFR NPT = 23,558 atoms
Typical Production MD NPT
&cntrl
ntx=5, irest=1,
ntc=2, ntf=2,
nstlim=10000,
ntpr=1000, ntwx=1000,
ntwr=10000,
dt=0.002, cut=8.,
ntt=1, tautp=10.0,
temp0=300.0,
ntb=2, ntp=1, taup=10.0,
ioutfm=1,
/
My system is smaller and my CPUs are better. It confused me that my speed should be much slower. Should I define the memory or something?
Need your help.
Cao Yang,
Zhejiang University, China
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri May 11 2012 - 02:00:04 PDT