For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Mojo function

split_smem

def split_smem[first_size: Int, second_size: Int, first_dtype: DType, second_dtype: DType](tensor: TileTensor[Storage=tensor.Storage, address_space=AddressSpace.SHARED, linear_idx_type=tensor.linear_idx_type, element_size=tensor.element_size]) -> Tuple[TileTensor[first_dtype, Layout[*?, *?], MutAnyOrigin, address_space=AddressSpace.SHARED], TileTensor[second_dtype, Layout[*?, *?], MutAnyOrigin, address_space=AddressSpace.SHARED]]

Split a shared memory tensor into two TileTensors at the boundary of first_size elements.

TMA only uses .ptr — flat row_major layout avoids needing InternalLayout equivalents of swizzled layouts.

Returns:

Tuple[TileTensor[first_dtype, Layout[*?, *?], MutAnyOrigin, address_space=AddressSpace.SHARED], TileTensor[second_dtype, Layout[*?, *?], MutAnyOrigin, address_space=AddressSpace.SHARED]]

Was this page helpful?

Thank you! We'll create more content like this.

Thank you for helping us improve!

😔 What went wrong?

Some code doesn’t work

It includes inaccurate information

It's missing information I need

It was difficult to understand

Other

Submit