v?add

Perform element-wise addition of vectors. The sum is also a vector.

Interface Definition

C interface:

void vsadd(const int len, const float* src1, const float* src2, float* dst);

void vdadd(const int len, const double* src1, const double* src2, double* dst);

void vcadd(const int len, const float complex *src1, const float complex *src2, float complex *dst);

void vzadd(const int len, const double complex *src1, const double complex *src2, double complex *dst);

void vhadd(const int len, const __fp16 *src1, const __fp16 *src2, __fp16 *dst);

void vchadd(const int len, const complex_fp16 *src1, const complex_fp16 *src2, complex_fp16 *dst);

Fortran interface:

CALL VSADD(LEN, SRC1, SRC2, DST);

CALL VDADD(LEN, SRC1, SRC2, DST);

CALL VZADD(LEN, SRC1, SRC2, DST);

Parameters

Parameter	Type	Description	Input/Output
len	Integer	Number of elements in the input vector. If len ≤ 0, an invalid parameter error is reported and the function returns.	Input
src1	Single-precision floating-point type for vsadd Double-precision floating-point type for vdadd Single-precision floating-point complex type for vcadd Double-precision floating-point complex type for vzadd Half-precision floating-point type for vhadd Half-precision floating-point complex type for vchadd	Input vector src1 with length len. If the pointer is null, a null pointer error is reported and the function returns.	Input
src2	Single-precision floating-point type for vsadd Double-precision floating-point type for vdadd Single-precision floating-point complex type for vcadd Double-precision floating-point complex type for vzadd Half-precision floating-point type for vhadd Half-precision floating-point complex type for vchadd	Input vector src2 with length len. If the pointer is null, a null pointer error is reported and the function returns.	Input
dst	Single-precision floating-point type for vsadd Double-precision floating-point type for vdadd Single-precision floating-point complex type for vcadd Double-precision floating-point complex type for vzadd Half-precision floating-point type for vhadd Half-precision floating-point complex type for vchadd	Output vector dst with length len. If the pointer is null, a null pointer error is reported and the function returns.	Output

Return Value

The sum of two numbers is returned for each operation value. The value range is (-INF, +INF).
For vsadd and vdadd, special output values are listed in the following table.
Input Value 1 (src1)

Input Value 2 (src2)

Output Value (dst)

+inf

-inf

nan

-inf

-inf

-inf

+inf

+inf

+inf

-inf

+inf

nan

nan

nan

nan
For vcadd and vzadd, special output values are listed in vcadd&vzadd.xlsx.

Input Value 1 (src1)	Input Value 2 (src2)	Output Value (dst)
+inf	-inf	nan
-inf	-inf	-inf
+inf	+inf	+inf
-inf	+inf	nan
nan	nan	nan

Dependencies

C: "kvml.h"

Examples

C interface:

    int i, len = 4; 
    float src1[len] = {0.0f, 0.5f, 1.5f, 5.5f}; 
    float src2[len] = {0.3f, 0.7f, 2.5f, 10.0f}; 
    float* dst = (float*)malloc(sizeof(float) * len); 
    if (dst == NULL) { 
        printf("Malloc Failed!\n"); 
        return 0; 
    } 
    vsadd(len, src1, src2, dst); 
    /** 
     *  Output dst: 
     *     0.3  1.2  4.0  15.5 
     * 
     */

Fortran interface:

    INTEGER :: LEN = 4 
    REAL(4) SRC1(4)  
    REAL(4) SRC2(4) 
    REAL(4) DST(4) 
    DATA SRC1 /1.2, 2.5, -2.4, 3.6/ 
    DATA SRC2 /8.6, -2.5, -12.4, 32.6/ 
    CALL VSADD(LEN, SRC1, SRC2, DST) 
    ! 
    ! OUTPUT DST 
    !     9.800000191, 0.000000000, -14.799999237, 36.199996948 
    !

Parent topic: Function Syntax