IMPORTANT: To view this page as Markdown, append `.md` to the URL (e.g. /get-started.md). For the complete documentation index, see llms.txt.
Skip to main content
For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Mojo function

padded_index_mode_kernel

def padded_index_mode_kernel[mode: StringSpan[ImmStaticOrigin], InputLayoutType: TensorLayout, input_origin: ImmOrigin, OutputLayoutType: TensorLayout, output_origin: MutOrigin, dtype: DType, rank: Int, InputEngine: TensorEngine, OutputEngine: TensorEngine](input_tensor: TileTensor[dtype, InputLayoutType, input_origin, Engine=InputEngine], output_tensor: TileTensor[dtype, OutputLayoutType, output_origin, Engine=OutputEngine], pre_pads: StaticTuple[Int32, rank], rows_per_sm: Int32, total_rows: Int32, row_length: Int32)

GPU kernel shared by the reflect and edge (repeat) pad modes.

Rows of the output are split across SMs and the last axis across threads. Axes 0..rank-2 of the source coordinate are computed once per row via the shared _pad_src_index mapping; the last axis is mapped per element.

Was this page helpful?