SIMD or not SIMD - cross platform

Question

I need some idea how to write a C++ cross platform implementation of a few parallelizable problems in a way so I can take advantage of SIMD (SSE, SPU, etc) if available. As well as I want to be able at run time to switch between SIMD and not SIMD.

How would you suggest me to approach this problem? (Of course I don't want to implement the problem multiple times for all possible options)

I can see how this might not be very easy task with C++ but I believe that I'm missing something. So far my idea looks like this... A class cStream will be array of a single field. Using multiple cStreams I can achieve SoA (Structure of Arrays). Then using a few Functors I can fake Lambda function that I need to be executed over the whole cStream.

// just for example I'm not expecting this code to compile
cStream a; // something like float[1024]
cStream b;
cStream c;

void Foo()
{
    for_each(
        AssignSIMD(c, MulSIMD(AddSIMD(a, b), a)));
}

Where for_each will be responsible for incrementing the current pointer of the streams as well as inlining the functors' body with SIMD and without SIMD.

something like so:

// just for example I'm not expecting this code to compile
for_each(functor<T> f)
{
#ifdef USE_SIMD
    if (simdEnabled)
        real_for_each(f<true>()); // true means use SIMD
    else
#endif
        real_for_each(f<false>());
}

Notice that if the SIMD is enabled is checked once and that the loop is around the main functor.

Answer 1

You might want to look at the source for the MacSTL library for some ideas in this area: www.pixelglow.com/macstl/