Deeploy.TilingExtension.TilingCodegen.padStride

Deeploy.TilingExtension.TilingCodegen.padStride(stride: Tuple[int, ...], rank: int, paddingStride: int) → Tuple[int, ...]

Pad a stride tuple to a target rank by prepending a specified stride value.

Extends a stride tuple to a higher dimensionality by adding leading stride values. This is useful for maintaining consistent stride calculations across different tensor ranks.

Parameters:

• stride (Tuple[int, ...]) – The original stride tuple to pad.

• rank (int) – The target rank (number of dimensions) for the padded stride.

• paddingStride (int) – The stride value to use for padding (prepended dimensions).

Returns:

The padded stride tuple with leading padding stride values.

Return type:

Tuple[int, …]

Raises:

AssertionError – If the target rank is smaller than the current stride’s rank.

Examples

>>> padStride((4, 1), 4, 16)
(16, 16, 4, 1)
>>> padStride((1,), 3, 8)
(8, 8, 1)