Problems with JOCL version of Dijkstra

dijkstra.cl

__kernel void dijkstra_initialize(
	__global int *maskArray,
	__global int *costArray,
	__global int *updatingCostArray,
	int sourceVertex,
	int vertexCount
) {
    // access thread id
    int tid = get_global_id(0);

	if (sourceVertex == tid) {
	    maskArray[tid] = 1;
	    costArray[tid] = 0;
	    updatingCostArray[tid] = 0;
	} else {
	    maskArray[tid] = 0;
	    costArray[tid] = INT_MAX;
	    updatingCostArray[tid] = INT_MAX;
	}
}

Dijkstra.java

program = clCreateProgramWithSource(context, 1,
		new String[] { SOURCE }, null, null);

clBuildProgram(program, 0, null, null, null, null);

final int vertexCount = graph.getVertexCount();
final int edgeCount = graph.getEdgeCount();

final int[] vertexArray = graph.getVertexArray();
final int[] edgeArray = graph.getEdgeArray();
final int[] weightArray = graph.getWeightArray();

final int[] maskArray = new int[vertexCount];
final int[] costArray = new int[vertexCount];
final int[] updatingCostArray = new int[vertexCount];

final Pointer vertexArrayPointer = Pointer.to(vertexArray);
final Pointer edgeArrayPointer = Pointer.to(edgeArray);
final Pointer weightArrayPointer = Pointer.to(weightArray);

final Pointer maskArrayPointer = Pointer.to(maskArray);
final Pointer costArrayPointer = Pointer.to(costArray);
final Pointer updatingCostArrayPointer = Pointer
		.to(updatingCostArray);

memObject = new cl_mem[6];
memObject[MEM_OBJECT_VERTEX_ARRAY] = clCreateBuffer(context,
		CL_MEM_READ_ONLY | CL_MEM_COPY_HOST_PTR, Sizeof.cl_int
				* vertexCount, vertexArrayPointer, null);
memObject[MEM_OBJECT_EDGE_ARRAY] = clCreateBuffer(context,
		CL_MEM_READ_ONLY | CL_MEM_COPY_HOST_PTR, Sizeof.cl_int
				* edgeCount, edgeArrayPointer, null);
memObject[MEM_OBJECT_WEIGHT_ARRAY] = clCreateBuffer(context,
		CL_MEM_READ_ONLY | CL_MEM_COPY_HOST_PTR, Sizeof.cl_int
				* edgeCount, weightArrayPointer, null);

memObject[MEM_OBJECT_MASK_ARRAY] = clCreateBuffer(context,
		CL_MEM_READ_WRITE, Sizeof.cl_int * vertexCount, null, null);
memObject[MEM_OBJECT_COST_ARRAY] = clCreateBuffer(context,
		CL_MEM_READ_WRITE, Sizeof.cl_int * vertexCount, null, null);
memObject[MEM_OBJECT_UPDATING_COST_ARRAY] = clCreateBuffer(context,
		CL_MEM_READ_WRITE, Sizeof.cl_int * vertexCount, null, null);

intializationKernel = clCreateKernel(program,
		KERNEL_INITIALIZATION, null);
sssp1Kernel = clCreateKernel(program, KERNEL_SSSP_1, null);
sssp2Kernel = clCreateKernel(program, KERNEL_SSSP_2, null);

// Initialization
//
// __global int *maskArray,
clSetKernelArg(intializationKernel, 0, Sizeof.cl_mem,
		Pointer.to(memObject[MEM_OBJECT_MASK_ARRAY]));
// __global float *costArray,
clSetKernelArg(intializationKernel, 1, Sizeof.cl_mem,
		Pointer.to(memObject[MEM_OBJECT_COST_ARRAY]));
// __global float *updatingCostArray,
clSetKernelArg(intializationKernel, 2, Sizeof.cl_mem,
		Pointer.to(memObject[MEM_OBJECT_UPDATING_COST_ARRAY]));
// int sourceVertexId
clSetKernelArg(intializationKernel, 3, Sizeof.cl_int,
		Pointer.to(new int[] { sourceVertexId }));
// int vertexCount
clSetKernelArg(intializationKernel, 4, Sizeof.cl_int,
		Pointer.to(new int[] { vertexCount }));

final long[] globalWorkSize = new long[] { vertexCount };

clEnqueueNDRangeKernel(queue, intializationKernel, DEFAULT_WORK_DIMENSION, null,
		globalWorkSize, DEFAULT_LOCAL_WORKSIZE, 0,
		null, null);

// DEBUG:START
// read mask array
clEnqueueReadBuffer(queue, memObject[MEM_OBJECT_MASK_ARRAY],
		CL_TRUE, 0, Sizeof.cl_int * vertexCount, maskArrayPointer, 0, null,
null);

System.out.println("Init");
System.out.println("mask: " + Arrays.toString(maskArray));
System.out.println("upco: " + Arrays.toString(updatingCostArray));
System.out.println("cost: " + Arrays.toString(costArray));
System.out.println("----");
// DEBUG:END

System.out

Init
mask: [-13071675, -13071675, -13071675, -13071675]
upco: [0, 0, 0, 0]
cost: [0, 0, 0, 0]
----