CDist (CPU)
===========

* Domain: ``com.microsoft``
* Device: ``CPU``
* Kernel: ``onnx_light_cpu::CDist``
* Supported types: ``FLOAT``, ``DOUBLE``
* Opset bounds: since opset 1

Operator support
----------------

.. list-table::
   :header-rows: 1
   :widths: 30 70

   * - Capability
     - Implementation
   * - Shape inference
     - :cpp:func:`onnx_light_cpu::ComputeShapeCDist`
   * - Peak memory
     - :cpp:func:`onnx_light_cpu::ComputePeakMemoryCDist`
   * - Fusion patterns
     - :cpp:class:`onnx_light_cpu::CDistFusionPattern`
   * - Gradient
     - :cpp:func:`onnx_light_cpu::RegisterCustomOperatorGradients`

LightOpSchema
~~~~~~~~~~~~~

**Version:** 1

Computes the pairwise Euclidean or squared Euclidean distances between the rows of two rank-2 tensors.

**Inputs**

- **A** (``T``): A rank-2 tensor with shape (M, N).
- **B** (``T``): A rank-2 tensor with shape (K, N).

**Outputs**

- **C** (``T``): The pairwise distances with shape (M, K).

**Attributes**

- **metric** (``STRING``): Distance metric: sqeuclidean or euclidean. (optional, default: ``sqeuclidean``)

**Type constraints**

- **T**: Constrain inputs and output to float or double tensors. Allowed types: tensor(double), tensor(float).