Der Amber users,
I've been dealing lately with a protein-dna complex, in which I want to compare the dna behavior in the presence and absence of an experimentally determined (via FRET) distance restraint.
Since, since last month, COM restraints are available in pmemd.cuda, I decided to give them a try.
My results are not really encouraging...
When using the following control file:
&cntrl
imin=0,irest=0,ntx=7,ioutfm=1,nmropt=1
ntb=2,pres0=1.0,ntp=1,barostat=2
taup=2.0,
ntr=1,ntc=2,ntf=2,
tempi=300.0,temp0=300.0,ntt=3,ig=-1,gamma_ln=5
nstlim=1000000,dt=0.002,
ntpr=10000,ntwx=10000,ntwr=100000
cut=8
/
&wt type='DUMPFREQ', istep1=2 /
&wt type='REST', istep1=0,istep2=1000000,value1=0.1,
value2=1, /
&wt type='END' /
DISANG=res.RST
DUMPAVE=bent_res_immo_min5.rst LISTIN=POUT
LISTOUT=POUT
END
With the following restraint file:
&rst iat= -1,-1, r1= 49.8, r2= 50.8, r3=55.76, r4=56.76, rk2=1, rk3=0.1, ir6=1, ialtd=0, igr1=1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63, igr2=579,580,581,582,583,584,585,586,587,588,589,590,591,592,593,594,595,596,597,598,599,600,601,602,603,604,605,606,607,608,609,610,611,612,613,614,615,616,617,618,619,620,621,622,623,624,625,626,627,628,629,630,631,632,633,634,635,636,637 /
I consistently get ~20 setps, after which the following error appears:
Error: an illegal memory access was encountered launching kernel kClearForces
cudaFree GpuBuffer::Deallocate failed an illegal memory access was encountered
I've tried it on GTX780 and on K20 cards, with the same results.
If I run the same simulation on CPUs, or if I modify the restraint file to present an atom based restraint, instead of a center of mass based one (i.e. i delete igr1 and igr2, and substitute iat= -1,-1, for iat= 1,579) and run it on either of the GPUs, the simulation can be run successfully.
I am not sure if my input files are bad, or if there might be any bugs on the COM implementation, but just thought it might be useful to post it here, and get some comments.
Thank you very much,
Manuel
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Feb 17 2015 - 01:30:02 PST
