for(int i=0;i<16384;i++)
         array[i]=expensive_function(array[i]);

     // use all GPUs and CPU at the same time
     var numberCruncher = new ClNumberCruncher(AcceleratorType.GPU|AcceleratorType.CPU,
    @"__kernel void acceleratedLoop(__global float *a)
       {
            int threadId=get_global_id(0);
            a[threadId]=pow(tanh(sqrt(cos(sin(a[threadId])))),0.3f);
       }");

    ClArray<float> buffer = array;
    buffer.compute(numberCruncher,1,"acceleratedLoop",16384);
    // now array has computed values by 16384 workitems on different devices such as 
    // gpus cpus igpus and fpgas 

https://www.codeproject.com/Articles/1181213/Easy-OpenCL-Multiple-Device-Load-Balancing-and-Pip

if you want to build the source on your computer yourself and to read a detailed wiki:

https://github.com/tugrul512bit/Cekirdekler/wiki