Re: AMBER: iwork array for nmr

From: Guanglei Cui <cuigl.csb.sunysb.edu>
Date: Mon, 14 Nov 2005 18:17:31 -0500

Hi, Peter

I've done something similar. To allocate more memory in WORK and IWORK
arrays, basically what I did is increase intreq and/or irlreq, which
you've done, and provided a pointer to the beginning of the newly
allocated array (search "JXPK require.*of storage"). I don't remember I
needed to create an array myself. But I may have misunderstood your
description.

You do remind me of b_nmri, which I didn't increase. My code works for
single CPU, now with b_nmri increased it should work correctly for MPI
too. I'll test it. Thank you.

Regards,

Guanglei

Peter Varnai wrote:
> Dear Amber-list,
>
> I would like to read a 'new' integer flag (related to a restraint)
> from a noe file and use it later in nmrnrg.f to call my subroutine.
>
> I used the exisiting 'ir6' as an example.
>
> I modified restal.f and nmrred.f to recognize 'new', created an array
> where this new(i) is stored (this works well) and then I try to pass
> back to nmrcal.
>
> In nmrcal, I added a maxrst size to iwork array, added 'new' to
> nmcloci common, changed b_nmri=52, in mpi_bcast I increased iwork
> size, intreq now equals to 14*maxrst + 5*maxwt + 2*maxgrp but iwork
> array is not holding the new(i) at all.
>
> If anyone knows how to solve this problem that would be a great.
> Besides, I am not sure how intreq or nmcloci is used in amber.
>
> Thanks,
> Peter
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Mon Nov 14 2005 - 23:53:00 PST
Custom Search