public void Run(ComputeContext context, TextWriter log)
{
try
{
ComputeProgram program = new ComputeProgram(context, kernelSources);
program.Build(null, null, null, IntPtr.Zero);
log.WriteLine("Program successfully built.");
ICollection<ComputeKernel> kernels = program.CreateAllKernels();
log.WriteLine("Kernels successfully created.");
// cleanup kernels
foreach (ComputeKernel kernel in kernels)
{
kernel.Dispose();
}
kernels.Clear();
// cleanup program
program.Dispose();
}
catch (Exception e)
{
log.WriteLine(e.ToString());
}
}
public void Run(ComputeContext context, TextWriter log)
{
this.log = log;
try
{
program = new ComputeProgram(context, clSource);
program.Build(null, null, notify, IntPtr.Zero);
}
catch (Exception e)
{
log.WriteLine(e.ToString());
}
// cleanup program
program.Dispose();
}
public static void Test()
{
string source = File.ReadAllText("MonteCarloSimulate.cl");
//Choose Device
ComputePlatform platform = ComputePlatform.Platforms[0];
ComputeDevice device = platform.QueryDevices()[0];
ComputeContextPropertyList properties =
new ComputeContextPropertyList(platform);
//Setup of stuff on our side
ComputeContext context = new ComputeContext(ComputeDeviceTypes.All,
properties, null, IntPtr.Zero);
//Build the program, which gets us the kernel
ComputeProgram program = new ComputeProgram(context, source);
program.Build(null, null, null, IntPtr.Zero);
//can use notify as the 3rd command... if you want this to be non-blocking
ComputeKernel kernel = program.CreateKernel("MonteCarloSimulate");
//Create arguments
int sideSize = 4096;
int[] inMatrixA = new int[sideSize * sideSize];
int[] inMatrixB = new int[sideSize * sideSize];
int[] outMatrixC = new int[sideSize * sideSize];
Random random = new Random((int)DateTime.Now.Ticks);
if (sideSize <= 32)
for (int y = 0; y < sideSize; y++)
for (int x = 0; x < sideSize; x++)
{
inMatrixA[y * sideSize + x] = random.Next(3);
inMatrixB[y * sideSize + x] = random.Next(3);
outMatrixC[y * sideSize + x] = 0;
}
ComputeBuffer<int> bufferMatrixA = new ComputeBuffer<int>(context,
ComputeMemoryFlags.UseHostPointer, inMatrixA);
ComputeBuffer<int> bufferMatrixB = new ComputeBuffer<int>(context,
ComputeMemoryFlags.UseHostPointer, inMatrixB);
ComputeBuffer<int> bufferMatrixC = new ComputeBuffer<int>(context,
ComputeMemoryFlags.UseHostPointer, outMatrixC);
long localWorkSize = Math.Min(device.MaxComputeUnits, sideSize);
//Sets arguments
kernel.SetMemoryArgument(0, bufferMatrixA);
kernel.SetMemoryArgument(1, bufferMatrixB);
kernel.SetMemoryArgument(2, bufferMatrixC);
kernel.SetLocalArgument(3, sideSize * 2);
kernel.SetValueArgument<int>(4, sideSize);
//kernel.SetLocalArgument(1, localWorkSize);
string offset = " ";
for (int x = 0; x < sideSize; x++)
offset += " ";
if (sideSize <= 32)
for (int y = 0; y < sideSize; y++)
{
Console.Write(offset);
for (int x = 0; x < sideSize; x++)
Console.Write(inMatrixA[y * sideSize + x] + " ");
Console.WriteLine();
}
//Runs commands
ComputeCommandQueue commands = new ComputeCommandQueue(context,
context.Devices[0], ComputeCommandQueueFlags.None);
long executionTime = DateTime.Now.Ticks;
//Execute kernel
//globalWorkSize in increments of localWorkSize (max of device.MaxComputeUnits or kernel.GetWorkGroupSize())
commands.Execute(kernel, null,
new long[] { Math.Min(sideSize, 16), Math.Min(sideSize, 16) },
new long[] { localWorkSize, 1 }, null);
//globalWorkSize can be any size
//localWorkSize product much not be greater than device.MaxComputeUnits
//and it must not be greater than kernel.GetWorkGroupSize()
//ESSENTIALLY, the program iterates through globalWorkSize
//in increments of localWorkSize. Both are multidimensional,
//but this just saves us the time of doing that
//(1 dimension can be put to multiple if the max dimension lengths
//are known very easily with remainder).
//Also, you should probably use this
//kernel.GetPreferredWorkGroupSizeMultiple(device);
commands.Finish();
commands.ReadFromBuffer(bufferMatrixC, ref outMatrixC, true, null);
commands.Finish();
executionTime = DateTime.Now.Ticks - executionTime;
GC.Collect();
program.Dispose();
Console.WriteLine();
if (sideSize <= 32)
for (int y = 0; y < sideSize; y++)
{
for (int x = 0; x < sideSize; x++)
Console.Write(inMatrixB[y * sideSize + x] + " ");
Console.Write(" ");
for (int x = 0; x < sideSize; x++)
Console.Write(outMatrixC[y * sideSize + x] + " ");
Console.WriteLine();
}
int testY = random.Next(sideSize);
int testX = random.Next(sideSize);
int sum = 0;
for (int q = 0; q < sideSize; q++)
sum += inMatrixA[q * sideSize + testX] *
inMatrixB[testY * sideSize + q];
Console.WriteLine(sum == outMatrixC[testY * sideSize + testX]);
Console.WriteLine(executionTime / 10000.0);
}
public void Run(ComputeContext context, TextWriter log)
{
try
{
// Create the arrays and fill them with random data.
int count = 10;
float[] arrA = new float[count];
float[] arrB = new float[count];
float[] arrC = new float[count];
Random rand = new Random();
for (int i = 0; i < count; i++)
{
arrA[i] = (float)(rand.NextDouble() * 100);
arrB[i] = (float)(rand.NextDouble() * 100);
}
// Create the input buffers and fill them with data from the arrays.
// Access modifiers should match those in a kernel.
// CopyHostPointer means the buffer should be filled with the data provided in the last argument.
ComputeBuffer<float> a = new ComputeBuffer<float>(context, ComputeMemoryFlags.ReadOnly | ComputeMemoryFlags.CopyHostPointer, arrA);
ComputeBuffer<float> b = new ComputeBuffer<float>(context, ComputeMemoryFlags.ReadOnly | ComputeMemoryFlags.CopyHostPointer, arrB);
// The output buffer doesn't need any data from the host. Only its size is specified (arrC.Length).
ComputeBuffer<float> c = new ComputeBuffer<float>(context, ComputeMemoryFlags.WriteOnly, arrC.Length);
// Create and build the opencl program.
program = new ComputeProgram(context, clProgramSource);
program.Build(null, null, null, IntPtr.Zero);
// Create the kernel function and set its arguments.
ComputeKernel kernel = program.CreateKernel("VectorAdd");
kernel.SetMemoryArgument(0, a);
kernel.SetMemoryArgument(1, b);
kernel.SetMemoryArgument(2, c);
// Create the event wait list. An event list is not really needed for this example but it is important to see how it works.
// Note that events (like everything else) consume OpenCL resources and creating a lot of them may slow down execution.
// For this reason their use should be avoided if possible.
ComputeEventList eventList = new ComputeEventList();
// Create the command queue. This is used to control kernel execution and manage read/write/copy operations.
ComputeCommandQueue commands = new ComputeCommandQueue(context, context.Devices[0], ComputeCommandQueueFlags.None);
// Execute the kernel "count" times. After this call returns, "eventList" will contain an event associated with this command.
// If eventList == null or typeof(eventList) == ReadOnlyCollection<ComputeEventBase>, a new event will not be created.
commands.Execute(kernel, null, new long[] { count }, null, eventList);
// Read back the results. If the command-queue has out-of-order execution enabled (default is off), ReadFromBuffer
// will not execute until any previous events in eventList (in our case only eventList[0]) are marked as complete
// by OpenCL. By default the command-queue will execute the commands in the same order as they are issued from the host.
// eventList will contain two events after this method returns.
commands.ReadFromBuffer(c, ref arrC, false, eventList);
// A blocking "ReadFromBuffer" (if 3rd argument is true) will wait for itself and any previous commands
// in the command queue or eventList to finish execution. Otherwise an explicit wait for all the opencl commands
// to finish has to be issued before "arrC" can be used.
// This explicit synchronization can be achieved in two ways:
// 1) Wait for the events in the list to finish,
//eventList.Wait();
// 2) Or simply use
commands.Finish();
// Print the results to a log/console.
for (int i = 0; i < count; i++)
log.WriteLine("{0} + {1} = {2}", arrA[i], arrB[i], arrC[i]);
// cleanup commands
commands.Dispose();
// cleanup events
foreach (ComputeEventBase eventBase in eventList)
{
eventBase.Dispose();
}
eventList.Clear();
// cleanup kernel
kernel.Dispose();
// cleanup program
program.Dispose();
// cleanup buffers
a.Dispose();
b.Dispose();
c.Dispose();
}
catch (Exception e)
{
log.WriteLine(e.ToString());
}
}
