How can I use Three Dimensional Array

[silvercall]

Hi,
I want to use three dimensional array e.g. int[][][] then store an array on third dimension
int[][][] dim = new int[5][5][120]

dim[0][0][0] = 4;
dim[0][0][1] = 44;
dim[0][0][2] = 12;

after set the values I want to start parallel loop

    for (@Parallel int i = 0; i < result[i].length; i++) {
        for (int j = 0; j < result[i][j].length; j++) {
            
        }
    }

is this somehow possible with TornadoVm. I have analysed every possible types like matrix, vector but I couldn't find any solution to solve this issue.

Can someone please help me?

[jjfumero]

TornadoVM does not support explicit 2D and 3D Java ND-Arrays. Instead, TornadoVM offers 2D and 3D Panama segments (implemented in with flatten arrays). I don't think we have for int types, but we have for float:

TornadoVM/tornado-unittests/src/main/java/uk/ac/manchester/tornado/unittests/matrices/TestMatrixTypes.java

Lines 614 to 644 in e74c574

	public void testMatrix10() throws TornadoExecutionPlanException {
	final int N = 256;
	Matrix3DFloat matrixA = new Matrix3DFloat(N, N, N);
	Matrix3DFloat matrixB = new Matrix3DFloat(N, N, N);
	Random r = new Random();
	for (int i = 0; i < N; i++) {
	for (int j = 0; j < N; j++) {
	for (int k = 0; k < N; k++) {
	matrixA.set(i, j, k, r.nextFloat());
	}
	}
	}
	TaskGraph taskGraph = new TaskGraph("nnnnnnn") //
	.transferToDevice(DataTransferMode.FIRST_EXECUTION, matrixA, matrixB) //
	.task("t0", TestMatrixTypes::computeMatrixSum, matrixA, matrixB, N) //
	.transferToHost(DataTransferMode.EVERY_EXECUTION, matrixB);
	ImmutableTaskGraph immutableTaskGraph = taskGraph.snapshot();
	try (TornadoExecutionPlan executionPlan = new TornadoExecutionPlan(immutableTaskGraph)) {
	executionPlan.execute();
	}
	for (int i = 0; i < N; i++) {
	for (int j = 0; j < N; j++) {
	for (int k = 0; k < N; k++) {
	assertEquals(matrixA.get(i, j, k) + matrixA.get(i, j, k), matrixB.get(i, j, k), 0.01f);
	}
	}
	}
	}

We can add a 3D int.

[silvercall]

Hi @jjfumero ,
thank you very much for the quick answer.
I will try to solve the issue with panama segments