c++ - Is it possible to use SIMD instruction for replace?

Question

I have vector of int and I need to find and replace some elements with specific value. Both of them are the same.
For example: replace 4 to 8 for all elements.

I'm trying direct memory access in loop in c++. But it still to slow for me.

Update:
I'm working with OpenCV Mat object on x86:

for (int i = 0; i < labels.rows; ++i) {
    for (int j = 0; j < labels.cols; ++j) {
        int& label = labels.at<int>(i, j);
        if (label == oldValue) {
            label = newValue;
        }
    }
}

Mat.at() function just return value by pointer in release mode

template<typename _Tp> inline
_Tp& Mat::at(int i0, int i1)
{
    CV_DbgAssert(dims <= 2);
    CV_DbgAssert(data);
    CV_DbgAssert((unsigned)i0 < (unsigned)size.p[0]);
    CV_DbgAssert((unsigned)(i1 * DataType<_Tp>::channels) < (unsigned)(size.p[1] * channels()));
    CV_DbgAssert(CV_ELEM_SIZE1(traits::Depth<_Tp>::value) == elemSize1());
    return ((_Tp*)(data + step.p[0] * i0))[i1];
}

Answer 1

You didn't mention what architecture you're developing for, so it's impossible to tell you which intrinsics to use. Luckily your compiler should be able to auto-vectorize something like

for (int i = 0 ; i < N ; i++)
  foo[i] = (foo[i] == 4) ? 8 : foo[i];

Assuming your data is sufficiently aligned, with -mavx2 -O3 GCC will use vpcmpeqd and vpblendvb.