Binary Cube

Björn Reuber

Hello,

there are many situation where a cube is used to "flag" a certain on/off condition (like SBO, dynamic select, matrix cubes, etc).

In most cases people are using integer cubes for this (in B10 a integer cube was 2byte per cell, in b12 it is 8byte per cell).

From application point of view one bit (so true or false) is enough to store this kind of information, so compared to Single we would save 1/32 of the cube size and for integer/double 1/64 of usage in Ram and on Disk

Regards

Björn

Julien CARDON

@Björn Reuber it means that in B12, an integer cube is bigger than a single cube ?

In that case is there any advantage to use an integer cube ?

Björn Reuber

Hello @Julien CARDON ,

as always it depends on the use case.

	Size	min	max	Precision
Single	4byte	-3.4028235E+38	+3.4028235E+38.	~6-9 digits
Double	8byte	-1.79769313486232E+308	+1.79769313486232E+308	~15-17 digits
Integer	8byte	-9,223,372,036,854,775,808	9,223,372,036,854,770,000	ALL 19 digits

With Integer you can display data with an huger precision, as it is possible with a double.

In b10 (and previous) integer was a 4byte value so froom -32k to 32k, and I had severall cases where we had an overflow, cause customers where not aware of the limit

regards

Björn

Julien CARDON

ok thanks for the feedback.

I'm using integer cubes to store 0 & 1 value because I was thinking that it was lighter in memory.

Björn Reuber

it was in previous versions, but with the downsize, that there was an overflow at around 32k