Hola Carmen,
Perhaps you could try to use the hard disk to combine the accumulated
weighted sums after the last iteration.
Continue frm your last _optimiser.star with --force_converge , but
without --dont_combine_weights_via_disc
HTH,
Sjors
On 07/31/2017 09:46 AM, Carmen San Martin wrote:
> Dear all,
>
> I am running relion2 autorefine procedure with a large particle (box size
> 752, pixel 1.37A). GPU cannot cope with it, so I am using a 16 node, 512 GB
> RAM machine. Relion keeps crashing at the last iteration with the following
> error:
>
> Calculating gold-standard FSC ...
> idx= 0ori_idx= 5137194714 idx_start= 18446744072098624320 idx_stop=
> 18446744072769634320 packed_size= 671010000
> ERROR:
> MlWsumModel::unpack: idx != idx_stop-idx_start
> File: /usr/local/scipion/software/em/relion-2.0.4/src/ml_model.cpp line:
> 1487
> --------------------------------------------------------------------------
>
> This happens even if I use only 3 MPI processes. I have previously run
> relion 1.4 with the same box size in a machine with only 256 GB RAM and had
> no problems.
>
> I did a google search and found this:
> https://github.com/3dem/relion/issues/180
>
> My relion version seems to already have the change suggested in the github
> issue, but it still crashes. Does anybody have any update on this?
>
> Best,
>
> CarmenSM
>
> -------------------------------------------------------------
> Carmen San MartĂn, Ph. D.
> Centro Nacional de BiotecnologĂa (CNB-CSIC)
> Darwin, 3
> 28049-Madrid (SPAIN)
> Email: [log in to unmask]
> Phone: 34-91-5855450
> Fax: 34-91-5854506
> http://tinyurl.com/carmensanmartinlab
> https://sites.google.com/site/adenonet/
> ------------------------------------------------------------
>
--
Sjors Scheres
MRC Laboratory of Molecular Biology
Francis Crick Avenue, Cambridge Biomedical Campus
Cambridge CB2 0QH, U.K.
tel: +44 (0)1223 267061
http://www2.mrc-lmb.cam.ac.uk/groups/scheres
|