2.2.3 : Column Automatic Padding



Column padding is very usefull to deal with matrix or tensors when their number of columns is not a multiple of a vectorial register size. This padding depends on the type of data and on the CPU architecture. The autopadding keyword has to be used on the last dimension of the tensor :

Example : 
1
2
3
4
5
6
7

///Table of values
TableVertex{
	///Image with padding
	Tensor(float, nbSlice, nbPixel+autopadding) rawImage;
	///Table of the calibrated slice signal (in pe) (nbEvent, nbSlice, nbPixel)
	Tensor(float, nbPixel+autopadding) tabCalibratedSignal;
}
In this case, the number of columns (nbPixel) of rawImage and tabCalibratedSignal will be allocated with extra columns if the number of pixels (nbPixel) is not a multiple of the vectorial register size of the CPU.