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

ds_read_tr16_b64_warp

ds_read_tr16_b64_warp[mma_shape: IndexList[3]](tile: TileTensor[address_space=AddressSpace.SHARED, linear_idx_type=tile.linear_idx_type, element_size=tile.element_size]) -> SIMD[tile.dtype, 4]

Warp-level transposed LDS read distributing across 16-lane rows.

For 32x32x16 MMA: 2x2 row distribution over 8x32 tile. For 16x16x32 MMA: 4x1 row distribution over 16x16 tile.

Parameters:

• ​mma_shape (IndexList[3]): MMA instruction shape (M, N, K).

Args:

• ​tile (TileTensor[address_space=AddressSpace.SHARED, linear_idx_type=tile.linear_idx_type, element_size=tile.element_size]): A TileTensor in shared memory sized for the MMA shape.

Returns:

SIMD[tile.dtype, 4]: A SIMD[dtype, 4] vector with transposed data for one lane.

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