tilelang.intrinsics.mma_sp_layout
=================================

.. py:module:: tilelang.intrinsics.mma_sp_layout


Functions
---------

.. autoapisummary::

   tilelang.intrinsics.mma_sp_layout.shared_16x16_to_mma_sp_layout_sr_a
   tilelang.intrinsics.mma_sp_layout.shared_16x16_to_mma_sp_layout_sr_b
   tilelang.intrinsics.mma_sp_layout.shared_16x32_to_mma_sp_layout_sr_a
   tilelang.intrinsics.mma_sp_layout.shared_16x32_to_mma_sp_layout_sr_b
   tilelang.intrinsics.mma_sp_layout.shared_16x64_to_mma_sp_layout_sr_a
   tilelang.intrinsics.mma_sp_layout.shared_16x64_to_mma_sp_layout_sr_b
   tilelang.intrinsics.mma_sp_layout.mma_sp_load_a_32x4_to_shared_16x16_layout
   tilelang.intrinsics.mma_sp_layout.mma_sp_load_a_32x8_to_shared_16x32_layout
   tilelang.intrinsics.mma_sp_layout.mma_sp_load_a_32x16_to_shared_16x64_layout
   tilelang.intrinsics.mma_sp_layout.mma_sp_load_b_32x8_to_shared_16x16_layout
   tilelang.intrinsics.mma_sp_layout.mma_sp_load_b_32x16_to_shared_16x32_layout
   tilelang.intrinsics.mma_sp_layout.mma_sp_load_b_32x32_to_shared_16x64_layout
   tilelang.intrinsics.mma_sp_layout.get_logical_id_32bit
   tilelang.intrinsics.mma_sp_layout.metadata_8bit_load_32x4_to_shared_16x4_layout_32bit
   tilelang.intrinsics.mma_sp_layout.metadata_16bit_load_32x2_to_shared_16x2_layout_32bit
   tilelang.intrinsics.mma_sp_layout.metadata_8bit_load_32x4_to_shared_16x4_layout_16bit
   tilelang.intrinsics.mma_sp_layout.metadata_16bit_load_32x2_to_shared_16x2_layout_16bit
   tilelang.intrinsics.mma_sp_layout.get_logical_id_8bit
   tilelang.intrinsics.mma_sp_layout.metadata_8bit_load_32x4_to_shared_16x4_layout_8bit
   tilelang.intrinsics.mma_sp_layout.metadata_16bit_load_32x2_to_shared_16x4_layout_8bit
   tilelang.intrinsics.mma_sp_layout.metadata_32bit_load_32x1_to_shared_16x2_layout_8bit
   tilelang.intrinsics.mma_sp_layout.ldmatrix_trans_32x8_to_shared_16x16_layout
   tilelang.intrinsics.mma_sp_layout.ldmatrix_32x16_to_shared_32x16_layout
   tilelang.intrinsics.mma_sp_layout.ldmatrix_trans_32x16_to_shared_16x32_layout
   tilelang.intrinsics.mma_sp_layout.ldmatrix_trans_32x32_to_shared_shared_16x64_layout
   tilelang.intrinsics.mma_sp_layout.get_ldmatrix_offset_b


Module Contents
---------------

.. py:function:: shared_16x16_to_mma_sp_layout_sr_a(i, j)

.. py:function:: shared_16x16_to_mma_sp_layout_sr_b(i, j)

.. py:function:: shared_16x32_to_mma_sp_layout_sr_a(i, j)

.. py:function:: shared_16x32_to_mma_sp_layout_sr_b(i, j)

.. py:function:: shared_16x64_to_mma_sp_layout_sr_a(i, j)

.. py:function:: shared_16x64_to_mma_sp_layout_sr_b(i, j)

.. py:function:: mma_sp_load_a_32x4_to_shared_16x16_layout(thread_id, local_id)

.. py:function:: mma_sp_load_a_32x8_to_shared_16x32_layout(thread_id, local_id)

.. py:function:: mma_sp_load_a_32x16_to_shared_16x64_layout(thread_id, local_id)

.. py:function:: mma_sp_load_b_32x8_to_shared_16x16_layout(thread_id, local_id)

.. py:function:: mma_sp_load_b_32x16_to_shared_16x32_layout(thread_id, local_id)

.. py:function:: mma_sp_load_b_32x32_to_shared_16x64_layout(thread_id, local_id)

.. py:function:: get_logical_id_32bit(thread_id)

.. py:function:: metadata_8bit_load_32x4_to_shared_16x4_layout_32bit(thread_id, local_id)

.. py:function:: metadata_16bit_load_32x2_to_shared_16x2_layout_32bit(thread_id, local_id)

.. py:function:: metadata_8bit_load_32x4_to_shared_16x4_layout_16bit(thread_id, local_id)

.. py:function:: metadata_16bit_load_32x2_to_shared_16x2_layout_16bit(thread_id, local_id)

.. py:function:: get_logical_id_8bit(thread_id)

.. py:function:: metadata_8bit_load_32x4_to_shared_16x4_layout_8bit(thread_id, local_id)

.. py:function:: metadata_16bit_load_32x2_to_shared_16x4_layout_8bit(thread_id, local_id)

.. py:function:: metadata_32bit_load_32x1_to_shared_16x2_layout_8bit(thread_id, local_id)

.. py:function:: ldmatrix_trans_32x8_to_shared_16x16_layout(thread_id, local_id)

.. py:function:: ldmatrix_32x16_to_shared_32x16_layout(thread_id, local_id)

.. py:function:: ldmatrix_trans_32x16_to_shared_16x32_layout(thread_id, local_id)

.. py:function:: ldmatrix_trans_32x32_to_shared_shared_16x64_layout(thread_id, local_id)

.. py:function:: get_ldmatrix_offset_b(matrix, row_idx, col_idx, stride, dtype = 'float16', transposed = False)