Hi all,
I think I know the answer to this but want to check
what experts have to say anyway.
I have a 1-dimensional dataset in a file whose elements
are multi-dimensional arrays (say [32][256][256]). All
together this looks like one large 4-dimesional array
(almost). Is there a way to read subsets of those
3-dimensional arrays into memory without reading whole
array? (I guess not)
Hyperslabs allow me to select only 1 dimension for the
dataset itself, I guess what I want to do is to extend
the hyperslab to the elements of the dataset which
are arrays themselves. I could not see anything in
documentation saying this is possible or impossible,
but my guess is that selection does not extend inside
single elements. Is that so?
Andy
Hi Andy,
Hi all,
I think I know the answer to this but want to check
what experts have to say anyway.
I have a 1-dimensional dataset in a file whose elements
are multi-dimensional arrays (say [32][256][256]). All
together this looks like one large 4-dimesional array
(almost). Is there a way to read subsets of those
3-dimensional arrays into memory without reading whole
array? (I guess not)
No, not currently.
Hyperslabs allow me to select only 1 dimension for the
dataset itself, I guess what I want to do is to extend
the hyperslab to the elements of the dataset which
are arrays themselves. I could not see anything in
documentation saying this is possible or impossible,
but my guess is that selection does not extend inside
single elements. Is that so?
You are correct. Currently, the only mechanism for selecting less than an entire element of a dataset is the way a subset of fields in a compound datatype can be accessed during a dataset read/write operation.
Quincey
···
On Oct 6, 2010, at 12:17 PM, Salnikov, Andrei A. wrote:
