[Numpy-discussion] ANN: BLZ 0.6.1 has been released

Valentin Haenel valentin at haenel.co
Sun Jan 26 10:44:41 EST 2014

Hi Dinesh Vadhia,

* Dinesh Vadhia <dineshbvadhia at hotmail.com> [2014-01-26]:
> For me, "binary data" wrt arrays means that data values are [0|1].  Is this 
> what is meant in "The compression process is carried out internally by 
> Blosc, a high-performance compressor that is optimized for binary data." ?

I believe, the term 'binary data' in this context refers to numerical
data -- e.g. floats and ints -- in the sense that it is not ascii or
other text.

Blosc is especially well suited for this kind of data due to its
optional shuffle filter. This filter will re-organize the bytes in the
data that is to be compressed in order of significance. For this filter
to work, each data value must be composed of multiple bytes, e.g.
int64.  For data values that are composed of a single byte, e.g. int8 or
char, the filter does not work so well.

Hope that helps,


More information about the NumPy-Discussion mailing list