[Nexus] large Nexus files

[Campbell, Branton]

Using NXputslab to reference a single voxel seems easy enough.  Thanks.

If I further want to simultaneously create and write two very large
arrays into separate Nexus data items within the same group, can I have
more than one data item open at a time?  My interpretation of the API
documentation suggests that the answer is "NO".

Branton


> -----Original Message-----
> From: Peterson, Peter F. [mailto:petersonpf at ornl.gov]
> Sent: Monday, July 17, 2006 9:04 PM
> To: Campbell, Branton; nexus at anl.gov
> Subject: RE: [Nexus] large Nexus files
> 
> Branton,
> 
> The way to do this is by creating the SDS (NXmakedata), then making
> several calls to NXputslab which don't need to be in any order.
> Remember, a single number is a "rectangular slab", however
inefficient.
> The assumption made is that you know the total size of the SDS upon
> starting.
> 
> P^2
> 
> -----Original Message-----
> From: nexus-bounces at anl.gov [mailto:nexus-bounces at anl.gov] On Behalf
Of
> Campbell, Branton
> Sent: Monday, July 17, 2006 10:34 PM
> To: nexus at anl.gov
> Subject: [Nexus] large Nexus files
> 
> Dear Nexus users,
> 
> Suppose that I have a large (much bigger than computer memory) 3D
array
> that I must create and write to a single Nexus dataset.  Also assume
> that I must compute the array values in an apparently random order.
> Have the Nexus developers envisioned an efficient way to use the Nexus
> API routines to write these elements in random order (i.e. not in nice
> rectangular slabs)?  In the past, I have used direct-access files for
> this purpose.
> 
> Of course, I would be better off sticking to arrays that fit in
memory.
> But I envision situations where this will not be possible (thousands
of
> 8MB images used to reconstruct large contiguous volumes of reciprocal
> space), and where it will be very expensive to break the output array
up
> into slabs and calculate them one at a time.
> 
> Best wishes,
> Branton
> 
> ****************************************
> Branton J. Campbell
> Department of Physics & Astronomy
> Brigham Young University
> N261 ESC, BYU, Provo, UT 84602
> Tel: 801-422-5758 Fax: 801-422-0553
> Email: branton_campbell at byu.edu
> http://www.physics.byu.edu/faculty/campbell/
> 
> 
> _______________________________________________
> NeXus mailing list
> NeXus at anl.gov
> http://www.neutron.anl.gov/mailman/listinfo/nexus