OpenCL bindings for Java

NJOCLSample.java

import java.nio.FloatBuffer;

import com.github.rjeschke.neetutils.collections.Colls;
import com.github.rjeschke.njocl.Buffers;
import com.github.rjeschke.njocl.CL;
import com.github.rjeschke.njocl.CLException;
import com.github.rjeschke.njocl.CommandQueue;
import com.github.rjeschke.njocl.Context;
import com.github.rjeschke.njocl.Device;
import com.github.rjeschke.njocl.Kernel;
import com.github.rjeschke.njocl.MemObject;
import com.github.rjeschke.njocl.Platform;
import com.github.rjeschke.njocl.Program;
import com.github.rjeschke.njocl.SizeOf;

/**
 * The OpenCL 'Hello world' using NJOCL.
 * 
 * Ported from: http://www.jocl.org/samples/JOCLSample.java
 * 
 * @author rjeschke
 */
public class NJOCLSample
{
    private static String PROGRAM_SOURCE = 
            "__kernel void " + 
            "sampleKernel(__global const float *a," + 
            "             __global const float *b," + 
            "             __global float *c)" + "{" + 
            "    int gid = get_global_id(0);" + 
            "    c[gid] = a[gid] * b[gid];" + 
            "}";

    public static void main(final String[] args) throws CLException
    {
        final int N = 16777216;
        final FloatBuffer srcArrayA = Buffers.newFloats(N);
        final FloatBuffer srcArrayB = Buffers.newFloats(N);
        final FloatBuffer outArray = Buffers.newFloats(N);
        // Fill input array (buffer)
        for (int i = 0; i < N; i++)
        {
            srcArrayA.put(i, i);
            srcArrayB.put(i, i);
        }

        // Get a (the first) platform
        final Platform platform = CL.getPlatform();
        // Get the first CPU device, create Context, CommandQueue and Program
        try (final Device device = platform.getDevice(CL.DEVICE_TYPE_CPU);
             final Context context = device.createContext();
             final CommandQueue cmdQueue = context.createCommandQueue(device, 0);
             final Program program = context.createProgramWithSource(PROGRAM_SOURCE))
        {
            // build the program
            program.buildProgram(null);
            
            // Create the Kernel and allocate MemObjects
            try (final Kernel kernel = program.createKernel("sampleKernel");
                 final MemObject srcMemA = context.createBuffer(CL.MEM_READ_ONLY | CL.MEM_USE_HOST_PTR, srcArrayA);
                 final MemObject srcMemB = context.createBuffer(CL.MEM_READ_ONLY | CL.MEM_USE_HOST_PTR, srcArrayB);
                 final MemObject outMem = context.createBuffer(CL.MEM_WRITE_ONLY | CL.MEM_ALLOC_HOST_PTR, N * SizeOf.FLOAT))
            {
                // Set Kernel arguments
                kernel.setArg(0, srcMemA);
                kernel.setArg(1, srcMemB);
                kernel.setArg(2, outMem);

                // Enqueue the Kernel
                cmdQueue.enqueueNDRangeKernel(kernel, 1, null, Colls.array((long)N), Colls.array(device.getMaxWorkGroupSize()));
                // Read back the results (blocking)
                cmdQueue.enqueueReadBuffer(outMem, outArray);
            }
        }

        // Check if computation succeeded
        boolean passed = true;
        final float epsilon = 1e-7f;
        for (int i = 0; i < N; i++)
        {
            final float x = outArray.get(i);
            final float y = srcArrayA.get(i) * srcArrayB.get(i);
            final boolean epsilonEqual = Math.abs(x - y) <= epsilon * Math.abs(x);
            if (!epsilonEqual)
            {
                passed = false;
                break;
            }
        }
        System.out.println("Test " + (passed ? "PASSED" : "FAILED"));
    }
}