cudf::segmented_reduce_host_udf Struct Reference

The interface for host-based UDF implementation for segmented reduction context. More...

#include <host_udf.hpp>

Inheritance diagram for cudf::segmented_reduce_host_udf:

Public Member Functions
virtual std::unique_ptr< column >	operator() (column_view const &input, device_span< size_type const > offsets, data_type output_dtype, null_policy null_handling, std::optional< std::reference_wrapper< scalar const >> init, rmm::cuda_stream_view stream, rmm::device_async_resource_ref mr) const =0
	Perform segmented reduction operations. More...
Public Member Functions inherited from cudf::host_udf_base
virtual	~host_udf_base ()=default
	Default destructor.
virtual std::size_t	do_hash () const
	Computes hash value of the instance. More...
virtual bool	is_equal (host_udf_base const &other) const =0
	Compares two instances of the derived class for equality. More...
virtual std::unique_ptr< host_udf_base >	clone () const =0
	Clones the instance. More...

Detailed Description

The interface for host-based UDF implementation for segmented reduction context.

An implementation of host-based UDF for segmented reduction needs to be derived from this class. In addition to implementing the virtual functions declared in the base class host_udf_base, such derived class must also define the operator() function to perform segmented reduction.

Example:

struct my_udf_aggregation : cudf::segmented_reduce_host_udf {
  my_udf_aggregation() = default;
 
  [[nodiscard]] std::unique_ptr<column> operator()(
    column_view const& input,
    device_span<size_type const> offsets,
    data_type output_dtype,
    null_policy null_handling,
    std::optional<std::reference_wrapper<scalar const>> init,
    rmm::cuda_stream_view stream,
    rmm::device_async_resource_ref mr) const override
  {
    // Perform computation using the input data and return the result.
    // This is where the actual segmented reduction logic is implemented.
  }
 
  [[nodiscard]] bool is_equal(host_udf_base const& other) const override
  {
    // Check if the other object is also instance of this class.
    // If there are internal state variables, they may need to be checked for equality as well.
    return dynamic_cast<my_udf_aggregation const*>(&other) != nullptr;
  }
 
  [[nodiscard]] std::unique_ptr<host_udf_base> clone() const override
  {
    return std::make_unique<my_udf_aggregation>();
  }
};

Definition at line 189 of file host_udf.hpp.

Member Function Documentation

operator()()

virtual std::unique_ptr<column> cudf::segmented_reduce_host_udf::operator() ( column_view const &  input, device_span< size_type const >  offsets, data_type  output_dtype, null_policy  null_handling, std::optional< std::reference_wrapper< scalar const >>  init, rmm::cuda_stream_view  stream, rmm::device_async_resource_ref  mr  ) const

pure virtual

Perform segmented reduction operations.

Parameters

input	The input column for reduction
offsets	A list of offsets defining the segments for reduction
output_dtype	The data type for the final output column
null_handling	If INCLUDE then the reduction result is valid only if all elements in the segment are valid, and if EXCLUDE then the reduction result is valid if any element in the segment is valid
init	The initial value of the reduction
stream	The CUDA stream to use for any kernel launches
mr	Device memory resource to use for any allocations

Returns

The output result of the aggregation

---

The documentation for this struct was generated from the following file:

• host_udf.hpp