Cast
=================



对输入数组进行 **逐元素数据类型转换（Cast）**，将源类型的数据按 C 语言强制类型转换规则转换为目标类型后输出。

数学表达式为：

.. math::

    dst_i = (T_{dst}) \; src_i

输入：
    - **input** - 输入数据地址。
    - **length** - 计算长度（元素个数）。
    - **core_mask** - 核掩码（仅适用于共享存储版本）。

输出：
    - **output** - 转换后的结果数据地址。

支持平台：
    ``FT78NE``
    ``MT7004``

.. note::
    - FT78NE 支持 ``fp32``、``fp64``、``int8``、``int16``、``int32``、``cplx64``、``cplx128`` 类型之间的 Cast
    - MT7004 支持 ``fp16``、``fp32``、``int16``、``int32``、``cplx64`` 类型之间的 Cast
    - 转换规则遵循 C 语言显式类型转换语义，可能存在精度截断或溢出风险
    - 输入与输出张量长度必须一致

**共享存储版本:**

.. c:function:: void i8_cast_to_i8_s(int8_t* input, int8_t* output, int length, int core_mask)
.. c:function:: void i8_cast_to_i16_s(int8_t* input, int16_t* output, int length, int core_mask)
.. c:function:: void i8_cast_to_i32_s(int8_t* input, int32_t* output, int length, int core_mask)
.. c:function:: void i8_cast_to_fp_s(int8_t* input, float* output, int length, int core_mask)

.. c:function:: void fp_cast_to_fp16_s(float* input, float16* output, int length, int core_mask);
.. c:function:: void fp_cast_to_int32_s(float* input, int* output, int length, int core_mask);
.. c:function:: void fp_cast_to_int16_s(float* input, short* output, int length, int core_mask);
.. c:function:: void fp_cast_to_fp16_p(float* input, float16* output, int length);
.. c:function:: void fp_cast_to_int32_p(float* input, int* output, int length);
.. c:function:: void fp_cast_to_int16_p(float* input, short* output, int length);

.. c:function:: void i16_cast_to_i8_s(int16_t* input, int8_t* output, int length, int core_mask)
.. c:function:: void i16_cast_to_i16_s(int16_t* input, int16_t* output, int length, int core_mask)
.. c:function:: void i16_cast_to_i32_s(int16_t* input, int32_t* output, int length, int core_mask)
.. c:function:: void i16_cast_to_fp_s(int16_t* input, float* output, int length, int core_mask)

.. c:function:: void i32_cast_to_i8_s(int32_t* input, int8_t* output, int length, int core_mask)
.. c:function:: void i32_cast_to_i16_s(int32_t* input, int16_t* output, int length, int core_mask)
.. c:function:: void i32_cast_to_i32_s(int32_t* input, int32_t* output, int length, int core_mask)
.. c:function:: void i32_cast_to_fp_s(int32_t* input, float* output, int length, int core_mask)

.. c:function:: void fp_cast_to_fp_s(float* input, float* output, int length, int core_mask)
.. c:function:: void fp_cast_to_int8_s(float* input, int8_t* output, int length, int core_mask)
.. c:function:: void fp_cast_to_int16_s(float* input, int16_t* output, int length, int core_mask)
.. c:function:: void fp_cast_to_int32_s(float* input, int32_t* output, int length, int core_mask)

.. c:function:: void dp_cast_to_dp_s(double* input, double* output, int length, int core_mask)
.. c:function:: void dp_cast_to_fp_s(double* input, float* output, int length, int core_mask)

.. c:function:: void c64_cast_to_c64_s(cplx64* input, cplx64* output, int length, int core_mask)
.. c:function:: void c128_cast_to_c128_s(cplx128* input, cplx128* output, int length, int core_mask)

**C调用示例：**

.. code-block:: c
    :linenos:
    :emphasize-lines: 12

    // FT78NE 示例：int8 -> int16
    #include <stdio.h>
    #include <cast.h>

    int main(int argc, char* argv[]) {
        int8_t *input = (int8_t *)0xA0000000;
        int16_t *output = (int16_t *)0xC0000000;

        int length = 2048;
        int core_mask = 0xff;

        i8_cast_to_i16_s(input, output, length, core_mask);
        return 0;
    }


**私有存储版本:**

.. c:function:: void i8_cast_to_i8_p(int8_t* input, int8_t* output, int length)
.. c:function:: void i8_cast_to_i16_p(int8_t* input, int16_t* output, int length)
.. c:function:: void i8_cast_to_i32_p(int8_t* input, int32_t* output, int length)
.. c:function:: void i8_cast_to_fp_p(int8_t* input, float* output, int length)

.. c:function:: void i16_cast_to_i8_p(int16_t* input, int8_t* output, int length)
.. c:function:: void i16_cast_to_i16_p(int16_t* input, int16_t* output, int length)
.. c:function:: void i16_cast_to_i32_p(int16_t* input, int32_t* output, int length)
.. c:function:: void i16_cast_to_fp_p(int16_t* input, float* output, int length)

.. c:function:: void i32_cast_to_i8_p(int32_t* input, int8_t* output, int length)
.. c:function:: void i32_cast_to_i16_p(int32_t* input, int16_t* output, int length)
.. c:function:: void i32_cast_to_i32_p(int32_t* input, int32_t* output, int length)
.. c:function:: void i32_cast_to_fp_p(int32_t* input, float* output, int length)

.. c:function:: void fp_cast_to_fp_p(float* input, float* output, int length)
.. c:function:: void fp_cast_to_int8_p(float* input, int8_t* output, int length)
.. c:function:: void fp_cast_to_int16_p(float* input, int16_t* output, int length)
.. c:function:: void fp_cast_to_int32_p(float* input, int32_t* output, int length)

.. c:function:: void dp_cast_to_dp_p(double* input, double* output, int length)
.. c:function:: void dp_cast_to_fp_p(double* input, float* output, int length)

.. c:function:: void c64_cast_to_c64_p(cplx64* input, cplx64* output, int length)
.. c:function:: void c128_cast_to_c128_p(cplx128* input, cplx128* output, int length)


**C调用示例：**

.. code-block:: c
    :linenos:
    :emphasize-lines: 10

    // FT78NE 示例：int8 -> int16（私有存储）
    #include <stdio.h>
    #include <cast.h>

    int main(int argc, char* argv[]) {
        int8_t *input = (int8_t *)0x10810000;
        int16_t *output = (int16_t *)0x10820000;

        int length = 2048;
        i8_cast_to_i16_p(input, output, length);
        return 0;
    }
