Question

任何人都可以告诉我为什么输出（Anew）是：

Anew = 0.000000    
Anew = 2.000000    
Anew = 4.000000    
Anew = 6.000000    
Anew = 16.000000    
Anew = 20.000000    
Anew = 24.000000    
Anew = 28.000000

而不是：

Anew = 0.000000    
Anew = 2.000000    
Anew = 4.000000    
Anew = 6.000000    
Anew = 8.000000    
Anew = 10.000000    
Anew = 12.000000    
Anew = 14.000000

代码：

#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include <mkl.h>

int main(int argc, const char* argv[]) {
    int rows = 2, cols = 2, Layers = 2;
    int PerLayerElmts = rows * cols;

    float* A = malloc(PerLayerElmts * Layers * sizeof(*A));

    // create A matrix
    int ImagIdx;
    for (int n = 0; n < Layers; n++) {
        ImagIdx = 0;
        for (int i = 0; i < rows; i++) {
            for (int j = 0; j < cols; j++) {
                A[n * PerLayerElmts + ImagIdx] = n * PerLayerElmts + ImagIdx;
                ImagIdx++;
            }
        }
    }
    // print A matrix
    for (int n = 0; n < Layers; n++) {
        ImagIdx = 0;
        for (int i = 0; i < rows; i++) {
            for (int j = 0; j < cols; j++) {
                printf("\nA = %f", A[n * PerLayerElmts + ImagIdx]);
                ImagIdx++;
            }
        }
    }
    float scalar = 2.0;
    size_t AddressOffset = 0;
    for (int i = 0; i < Layers; i++, AddressOffset += PerLayerElmts) {
        // multiply A by scalar
        cblas_sscal(PerLayerElmts * Layers, scalar, A + AddressOffset, 1);
    }
    // print A matrix
    for (int n = 0; n < Layers; n++) {
        ImagIdx = 0;
        for (int i = 0; i < rows; i++) {
            for (int j = 0; j < cols; j++) {
                printf("\n\nAnew = %f", A[n * PerLayerElmts + ImagIdx]);
                ImagIdx++;
            }
        }
    }

    printf("\n");

    free(A);

    return 0;
}

我只是创建一个矩阵，然后使用cblas调用将每个元素乘以标量（2）。

我这样做，使用addressoffset作为图层数。

问题出在第二层，其中元素乘以4而不是2！

Answer 1

你打电话cblas_sscal的方式似乎不对。而不是

cblas_sscal( PerLayerElmts * Layers , scalar , A + AddressOffset , 1 );

我希望像

这样的东西

cblas_sscal( PerLayerElmts , scalar , A + AddressOffset , 1 );

因为你为每一层调用一次。

为什么我在数组的后半部分得到这个结果？

1 个答案: