[AMBER] Amber18 TI on K10 - any known problem?

From: Thomas Fox <thomas_fox.gmx.net>
Date: Wed, 13 Feb 2019 13:06:28 +0100


   Im having problems running TI calculations on old K10 GPUs. About 10% of my
   jobs produce after a while NaNs in the output. Before that, the run looks
   pretty normal. The bombs are not reproducible, when I run the same set of
   ligand perturbations multiple times, its always different perturbations and
   different lambda values that fails. The same systems ran fine multiple times
   with Amber16.

   My setup: Amber18, with all updates installed, RH SL6.8 with Cuda8.0. TI
   with softcoremasks, with/without HMR. I only observe the crashes on
   K10.G1.8GB or K10.G2.8GB, runs on K40m or K40c or GTX1080 cards seem to be

   Any idea how I could trace the source of the problem? Any known issues I
   should be aware of - e.g. some weird combination of input parameters that I
   use? For reference, I have attached my mdin file (and please no flames - I
   know that my input parameters do not reflect the current state of the art -
   but hints for improvement are certainly welcome :-).

   Thanks a lot,

   # system in water box, production at 300K NPT
      ntx=5, irest=1,
      ntpr=1000, ntwx=0, ntwe=0, ntwr=250000, ntwprt=0, ioutfm=1,
       ntf=1, ntb=2, vdw_cutoff=9.0, es_cutoff=9.0, ntp=1, pres0=1.0,
   comp=44.6, taup=1.0,
      barostat = 2,
      nstlim=2500000, dt=0.004,
      ntt=3, ig=-1, temp0=300.0, tautp=1.0, gamma_ln=3,
      ene_avg_sampling=1, logdvdl=1,
      icfe=1, clambda= 0.4,
      timask1=':1', timask2=':2',
      ifsc=1, scbeta=18,


      mbar_lambda=0.00, 0.05, 0.10, 0.15, 0.20, 0.25, 0.30, 0.35, 0.40, 0.45,
                   0.55, 0.60, 0.65, 0.70, 0.75, 0.80, 0.85, 0.90, 0.95, 1.0
AMBER mailing list
Received on Wed Feb 13 2019 - 04:30:02 PST
Custom Search