Haha, random code you say - what an interesting idea!

The funny thing is that this is still the simple stuff. OpenMP is so full of features...

But Nvidia's CUDA API for GPU Computing is more difficult,
and let's please not even talk about OpenCL, which is even more complicated...