/**
* Constructor initializes xyz and vxyz arrays.
*
* @param _bodies
*/
public NBodyKernel(Range _range) {
range = _range;
localStuff = new float[range.getLocalSize(0) * 3];
xyz = new float[range.getGlobalSize(0) * 3];
vxyz = new float[range.getGlobalSize(0) * 3];
final float maxDist = 20f;
for (int body = 0; body < (range.getGlobalSize(0) * 3); body += 3) {
final float theta = (float) (Math.random() * Math.PI * 2);
final float phi = (float) (Math.random() * Math.PI * 2);
final float radius = (float) (Math.random() * maxDist);
// get the 3D dimensional coordinates
xyz[body + 0] = (float) (radius * Math.cos(theta) * Math.sin(phi));
xyz[body + 1] = (float) (radius * Math.sin(theta) * Math.sin(phi));
xyz[body + 2] = (float) (radius * Math.cos(phi));
// divide into two 'spheres of bodies' by adjusting x
if ((body % 2) == 0) {
xyz[body + 0] += maxDist * 1.5;
} else {
xyz[body + 0] -= maxDist * 1.5;
}
}
setExplicit(true);
}
public static void main(String[] _args) {
final int size = 512;
final float[] a = new float[size];
final float[] b = new float[size];
for (int i = 0; i < size; i++) {
a[i] = (float) (Math.random() * 100);
b[i] = (float) (Math.random() * 100);
}
final float[] sum = new float[size];
Kernel kernel =
new Kernel() {
@Override
public void run() {
int gid = getGlobalId();
sum[gid] = a[gid] + b[gid];
}
};
kernel.execute(Range.create(512));
for (int i = 0; i < size; i++) {
System.out.printf("%6.2f + %6.2f = %8.2f\n", a[i], b[i], sum[i]);
}
kernel.dispose();
}
public static void main(String[] args) {
final int WIDTH = 1024 * 16;
final int HEIGHT = 1024 * 8;
final int BIN_SIZE = 128;
final int GROUP_SIZE = 128;
final int SUB_HISTOGRAM_COUNT = ((WIDTH * HEIGHT) / (GROUP_SIZE * BIN_SIZE));
final byte[] data = new byte[WIDTH * HEIGHT];
for (int i = 0; i < (WIDTH * HEIGHT); i++) {
data[i] = (byte) ((Math.random() * BIN_SIZE) / 2);
}
final byte[] sharedArray = new byte[GROUP_SIZE * BIN_SIZE];
final int[] binResult = new int[SUB_HISTOGRAM_COUNT * BIN_SIZE];
System.out.println("binResult size=" + binResult.length);
final int[] histo = new int[BIN_SIZE];
final int[] refHisto = new int[BIN_SIZE];
final Device device = KernelManager.instance().bestDevice();
if (device != null) {
System.out.println(((OpenCLDevice) device).getOpenCLPlatform().getName());
final Range rangeBinSize = device.createRange(BIN_SIZE);
final Range range = Range.create((WIDTH * HEIGHT) / BIN_SIZE, GROUP_SIZE);
if (device instanceof OpenCLDevice) {
final OpenCLDevice openclDevice = (OpenCLDevice) device;
final HistogramKernel histogram =
openclDevice.bind(
HistogramKernel.class,
Histogram.class
.getClassLoader()
.getResourceAsStream("com/amd/aparapi/sample/extension/HistogramKernel.cl"));
long start = System.nanoTime();
histogram
.begin() //
.put(data) //
.histogram256(range, data, sharedArray, binResult, BIN_SIZE) //
// by leaving binResult on the GPU we can save two 1Mb transfers
.bin256(rangeBinSize, histo, binResult, SUB_HISTOGRAM_COUNT) //
.get(histo) //
.end();
System.out.println("opencl " + ((System.nanoTime() - start) / 1000000));
start = System.nanoTime();
for (int i = 0; i < (WIDTH * HEIGHT); i++) {
refHisto[data[i]]++;
}
System.out.println("java " + ((System.nanoTime() - start) / 1000000));
for (int i = 0; i < 128; i++) {
if (refHisto[i] != histo[i]) {
System.out.println(i + " " + histo[i] + " " + refHisto[i]);
}
}
}
} else {
System.out.println("no GPU device");
}
}
/**
* Render all particles to the OpenGL context
*
* @param gl
*/
protected void render(GL2 gl) {
gl.glBegin(GL2.GL_QUADS);
for (int i = 0; i < (range.getGlobalSize(0) * 3); i += 3) {
gl.glTexCoord2f(0, 1);
gl.glVertex3f(xyz[i + 0], xyz[i + 1] + 1, xyz[i + 2]);
gl.glTexCoord2f(0, 0);
gl.glVertex3f(xyz[i + 0], xyz[i + 1], xyz[i + 2]);
gl.glTexCoord2f(1, 0);
gl.glVertex3f(xyz[i + 0] + 1, xyz[i + 1], xyz[i + 2]);
gl.glTexCoord2f(1, 1);
gl.glVertex3f(xyz[i + 0] + 1, xyz[i + 1] + 1, xyz[i + 2]);
}
gl.glEnd();
}
public static void main(String _args[]) {
final NBodyKernel kernel =
new NBodyKernel(Range.create(Integer.getInteger("bodies", 8192), 256));
final JFrame frame = new JFrame("NBody");
final JPanel panel = new JPanel(new BorderLayout());
final JPanel controlPanel = new JPanel(new FlowLayout());
panel.add(controlPanel, BorderLayout.SOUTH);
final JButton startButton = new JButton("Start");
startButton.addActionListener(
new ActionListener() {
@Override
public void actionPerformed(ActionEvent e) {
running = true;
startButton.setEnabled(false);
}
});
controlPanel.add(startButton);
controlPanel.add(new JLabel(kernel.getExecutionMode().toString()));
controlPanel.add(new JLabel(" Particles"));
controlPanel.add(new JTextField("" + kernel.range.getGlobalSize(0), 5));
controlPanel.add(new JLabel("FPS"));
final JTextField framesPerSecondTextField = new JTextField("0", 5);
controlPanel.add(framesPerSecondTextField);
controlPanel.add(new JLabel("Score("));
final JLabel miniLabel =
new JLabel("<html><small>calcs</small><hr/><small>µsec</small></html>");
controlPanel.add(miniLabel);
controlPanel.add(new JLabel(")"));
final JTextField positionUpdatesPerMicroSecondTextField = new JTextField("0", 5);
controlPanel.add(positionUpdatesPerMicroSecondTextField);
final GLCapabilities caps = new GLCapabilities(null);
caps.setDoubleBuffered(true);
caps.setHardwareAccelerated(true);
final GLCanvas canvas = new GLCanvas(caps);
final Dimension dimension =
new Dimension(Integer.getInteger("width", 742), Integer.getInteger("height", 742));
canvas.setPreferredSize(dimension);
canvas.addGLEventListener(
new GLEventListener() {
private double ratio;
private final float xeye = 0f;
private final float yeye = 0f;
private final float zeye = 100f;
private final float xat = 0f;
private final float yat = 0f;
private final float zat = 0f;
public final float zoomFactor = 1.0f;
private int frames;
private long last = System.currentTimeMillis();
@Override
public void dispose(GLAutoDrawable drawable) {}
@Override
public void display(GLAutoDrawable drawable) {
final GL2 gl = drawable.getGL().getGL2();
gl.glLoadIdentity();
gl.glClear(GL.GL_COLOR_BUFFER_BIT | GL.GL_DEPTH_BUFFER_BIT);
gl.glColor3f(1f, 1f, 1f);
final GLU glu = new GLU();
glu.gluPerspective(45f, ratio, 0f, 1000f);
glu.gluLookAt(xeye, yeye, zeye * zoomFactor, xat, yat, zat, 0f, 1f, 0f);
if (running) {
kernel.execute(kernel.range);
if (kernel.isExplicit()) {
kernel.get(kernel.xyz);
}
final List<ProfileInfo> profileInfo = kernel.getProfileInfo();
if ((profileInfo != null) && (profileInfo.size() > 0)) {
for (final ProfileInfo p : profileInfo) {
System.out.print(
" "
+ p.getType()
+ " "
+ p.getLabel()
+ ((p.getEnd() - p.getStart()) / 1000)
+ "us");
}
System.out.println();
}
}
kernel.render(gl);
final long now = System.currentTimeMillis();
final long time = now - last;
frames++;
if (time > 1000) { // We update the frames/sec every second
if (running) {
final float framesPerSecond = (frames * 1000.0f) / time;
final int updatesPerMicroSecond =
(int)
((framesPerSecond
* kernel.range.getGlobalSize(0)
* kernel.range.getGlobalSize(0))
/ 1000000);
framesPerSecondTextField.setText(String.format("%5.2f", framesPerSecond));
positionUpdatesPerMicroSecondTextField.setText(
String.format("%4d", updatesPerMicroSecond));
}
frames = 0;
last = now;
}
gl.glFlush();
}
@Override
public void init(GLAutoDrawable drawable) {
final GL2 gl = drawable.getGL().getGL2();
gl.glShadeModel(GLLightingFunc.GL_SMOOTH);
gl.glEnable(GL.GL_BLEND);
gl.glBlendFunc(GL.GL_SRC_ALPHA, GL.GL_ONE);
try {
final InputStream textureStream = Local.class.getResourceAsStream("particle.jpg");
final Texture texture = TextureIO.newTexture(textureStream, false, null);
texture.enable(gl);
} catch (final IOException e) {
e.printStackTrace();
} catch (final GLException e) {
e.printStackTrace();
}
}
@Override
public void reshape(GLAutoDrawable drawable, int x, int y, int _width, int _height) {
width = _width;
height = _height;
final GL2 gl = drawable.getGL().getGL2();
gl.glViewport(0, 0, width, height);
ratio = (double) width / (double) height;
}
});
panel.add(canvas, BorderLayout.CENTER);
frame.getContentPane().add(panel, BorderLayout.CENTER);
frame.setDefaultCloseOperation(WindowConstants.EXIT_ON_CLOSE);
frame.pack();
frame.setVisible(true);
final FPSAnimator animator = new FPSAnimator(canvas, 100);
animator.start();
}