APIs that Compilation Depends on

The following APIs are used only during operator compilation, and are not directly called during operator writing: You can view the API definitions in the ddk/include/inc/tensor_engine/cce_aicpu_3rd_party.h file under the installation directory of the DDK package.

• __aicpu__ void *aicpu_malloc(unsigned int size)

  This API is used to allocate memory. The input parameter size indicates the size of memory to be allocated. The pointer to memory allocation is returned.

• __aicpu__ void aicpu_free(void * ptr)

  This API is used to release memory. ptr is the memory pointer.

• __aicpu__ double log(double x)

  This API is used to obtain the log value of x (double type).

• __aicpu__ double exp(double x)

  This API is used to obtain the EXP value of x (double type).

• __aicpu__ double round(double x)

  This API is used to round off the value of x (double type).

• __aicpu__ double floor(double x)

  This API is used to round down the value of x (double type).

• __aicpu__ double ceil(double x);

  This API is used to round up the value of x (double type).

• __aicpu__ double trunc(double x);

  This API is used to obtain the truncation value of x (double type).

• __aicpu__ double sqrt(double x);

  This API is used to obtain the square value of x (double type).

• __aicpu__ float logf(float x)

  This API is used to obtain the log value of x (double type).

• __aicpu__ float expf(float x)

  This API is used to obtain the EXP value of x (float type).

• __aicpu__ float roundf(float x)

  This API is used to round off the value of x (float type).

• __aicpu__ float floorf(float x)

  This API is used to round down the value of x (float type).

• __aicpu__ float ceilf(float x);

  This API is used to round up the value of x (float type).

• __aicpu__ float truncf(float x);

  This API is used to obtain the truncation value of x (float type).

• __aicpu__ float sqrtf(float x);

  This API is used to obtain the square value of x (float type).