tilelang.cuda.op.gemm.gemm_mma_sm120
====================================

.. py:module:: tilelang.cuda.op.gemm.gemm_mma_sm120


Attributes
----------

.. autoapisummary::

   tilelang.cuda.op.gemm.gemm_mma_sm120.GEMM_INST_MMA_BLOCK_SCALED


Classes
-------

.. autoapisummary::

   tilelang.cuda.op.gemm.gemm_mma_sm120.GemmMMASm120BlockScaled


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

.. py:data:: GEMM_INST_MMA_BLOCK_SCALED
   :value: 'cuda.mma.blockscaled'


.. py:class:: GemmMMASm120BlockScaled

   Bases: :py:obj:`tilelang.cuda.op.gemm.gemm_mma.GemmMMA`


   SM120 warp-level block-scaled MMA lowering.


   .. py:attribute:: intrin_emitter_cls


   .. py:method:: infer_layout(target, thread_nums)


   .. py:method:: lower(layout_map, target, thread_bounds, thread_index, mbar_phase_expr = None)