Skip to main content

Mojo function

mma

mma[: DType, : DType, : IndexList[3], : Bool, : Int, : Int, : Int, : Int, : Int, : Swizzle, : Int, //, k_tile_idx: Int, swap_a_b: Bool, MMAType: AnyStruct[AMD_MMA[$0, $1, $2, $3, $4, $5, $6, $7, $8, $9, $10]]](a_tiles: MMATileBuffers[smem_layout, tensor_type, thread_layout, block_rows, warp_rows, stride, num_mmas, MMAType], b_tiles: MMATileBuffers[smem_layout, tensor_type, thread_layout, block_rows, warp_rows, stride, num_mmas, MMAType], c_reg_tile: LayoutTensor[dtype, layout, origin, address_space=address_space, element_layout=element_layout, layout_int_type=layout_int_type, linear_idx_type=linear_idx_type, masked=masked, alignment=alignment])

Was this page helpful?