tilelang.contrib.ptodsl.simt
============================

.. py:module:: tilelang.contrib.ptodsl.simt

.. autoapi-nested-parse::

   PTODSL tracing helpers for TileLang PTO SIMT codegen.



Functions
---------

.. autoapisummary::

   tilelang.contrib.ptodsl.simt.shuffle_vec
   tilelang.contrib.ptodsl.simt.vector_from_list
   tilelang.contrib.ptodsl.simt.vector_to_list
   tilelang.contrib.ptodsl.simt.store_vector_to_list
   tilelang.contrib.ptodsl.simt.vectorize_unary_f32x2
   tilelang.contrib.ptodsl.simt.vectorize_binary_f32x2
   tilelang.contrib.ptodsl.simt.vectorize_binary_fp8
   tilelang.contrib.ptodsl.simt.fp8_byte_load
   tilelang.contrib.ptodsl.simt.fp8_byte_store
   tilelang.contrib.ptodsl.simt.scalar_binary_fp8
   tilelang.contrib.ptodsl.simt.scalar_div
   tilelang.contrib.ptodsl.simt.scalar_rsqrt
   tilelang.contrib.ptodsl.simt.simt_allreduce_sum
   tilelang.contrib.ptodsl.simt.simt_allreduce_max
   tilelang.contrib.ptodsl.simt.simt_allreduce_min


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

.. py:function:: shuffle_vec(dtype, values, indices)

   Pick lanes by constant index from concatenated scalar/vector inputs.


.. py:function:: vector_from_list(dtype, values)

   Build a PTODSL builtin vector from scalar values in a Python sequence.


.. py:function:: vector_to_list(value)

   Extract all lanes of a PTODSL builtin vector into a Python list.


.. py:function:: store_vector_to_list(dst, offset, value)

   Store all lanes of a PTODSL vector into a Python-list local buffer.


.. py:function:: vectorize_unary_f32x2(op, value)

.. py:function:: vectorize_binary_f32x2(op, lhs, rhs)

.. py:function:: vectorize_binary_fp8(op, lhs, rhs, dtype)

   Compute FP8 vectors through the supported packed two-lane conversions.


.. py:function:: fp8_byte_load(buffer, index, dtype)

   Load one FP8 storage byte without materializing an unsupported scalar FP8.


.. py:function:: fp8_byte_store(value, buffer, index, dtype)

   Store one FP8 storage byte through its integer representation.


.. py:function:: scalar_binary_fp8(op, lhs, rhs, dtype)

   Use the packed FP8 conversion primitive for one logical element.


.. py:function:: scalar_div(lhs, rhs)

.. py:function:: scalar_rsqrt(value)

.. py:function:: simt_allreduce_sum(value, **kwargs)

.. py:function:: simt_allreduce_max(value, **kwargs)

.. py:function:: simt_allreduce_min(value, **kwargs)