Re: [scipion-users] particle polishing

[Josue G. B. <jos...@gm...>]

Dear Dmitry:
It seems that you have a problem with the RAM memory available in your
cluster. In version 1.0.1 the results of  cross-correlation+Optical Flow
(alignining your movies) are the same as particle polishing, and it is a
simpler way to process the particles. Unfortunately, particle polishing
demands  a huge of RAM memory and time, and depends of your particle
dimensions and the number of particles. The only  way to decrease the
amount of memory without change the dimensions of the images is decreasing
the number of MPI processes (until 1), but, of course, there exist a
minimum... In previous projects I needed a worstation (1node) with 512GB of
RAM to successfully execute particle polish with 3 MPI process (I dont
remember the box size, but no more than 400 pixels). Maybe you can try with
a compilation of relion in single precision.... Can you send me a log error
to see the problem?...
Best Regards, Josue

2016-12-08 16:24 GMT+01:00 Dmitry Semchonok <sem...@gm...>:

> Dear colleagues,
>
> I'm still on the stage of particle polishing (from the frames)
>
>
> What are on average *the setting* that you use for the
>
>
> 1) MPI
>
> 2) nodes/himem - ?
>
> What is on average
>
> *the memory per/core (in himem mode)? *I tried 20480; 40960 = error
>
> I have something around 800 000 frames
>
>
>
>
> Sincerely
> Dmitry
>
> ------------------------------------------------------------
> ------------------
> Developer Access Program for Intel Xeon Phi Processors
> Access to Intel Xeon Phi processor-based developer platforms.
> With one year of Intel Parallel Studio XE.
> Training and support from Colfax.
> Order your platform today.http://sdm.link/xeonphi
> _______________________________________________
> scipion-users mailing list
> sci...@li...
> https://lists.sourceforge.net/lists/listinfo/scipion-users
>
>