Thanks! I thought I'd tried this before with no success, but perhaps I
made a mistake.
Regards,
-jh-
On 12/15/18 9:28 AM, Alpay Burak Seven wrote:
> 4 mpi and gpu values of 0:1:2:3 work for me for 4 gpus (not commas or
> spaces).
>
> On Sat, Dec 15, 2018, 3:18 AM Takanori Nakane
> <[log in to unmask] <mailto:[log in to unmask]> wrote:
>
> Hi,
>
> RELION's MotionCor2 wrapper assumes one GPU per MPI process.
> If you have 4 GPUs on a machine, you should run 4 MPI processes
> (= 4 MotionCor2 processes), each using one GPU.
> Is assigning all 4 GPUs to one MotionCor2 faster than
> running 4 instances of MotionCor2?
>
> Best regards,
>
> Takanori Nakane
>
> On 2018/12/15 3:42, John Heumann wrote:
> > I realize Relion now has it's own implementation of motion
> correction, but for those using the motionCor2 wrapper, I wonder
> if it would be possible make a minor change which would allow
> using multiple GPUs from the gui using notation consistent withÂ
> that used elsewhere in Relion? Specifically, commas are typically
> used to separate GPUs (e.g, "0,1,2,3") for a single process in the
> Relion Gui. But this doesn't seem to work correctly for motionCor2
> 1.2.1. MotionCor2 exepcts GPUs separated by spaces and doesn't
> seem to accept commas. At least in my hands, it winds up using
> only a single GPU if commas are entered in the gui. What does seem
> to work, at least for me is entering the GPUs separated by spaces
> (e.g. "0 1 2 3", but without the quotations). The difference in
> speed is considerable obviously.
> >
> > Thanks for considering this!
> >
> >
> ########################################################################
> >
> > To unsubscribe from the CCPEM list, click the following link:
> > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
> >
>
> ########################################################################
>
> To unsubscribe from the CCPEM list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
>
>
> ------------------------------------------------------------------------
>
> To unsubscribe from the CCPEM list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
>
--
John M. Heumann
Department of Molecular, Cellular, and Developmental Biology
347 UCB, University of Colorado
Boulder, CO 80309-0347
########################################################################
To unsubscribe from the CCPEM list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1
|