/// <summary>
/// Print GPU for primary card
/// </summary>
static void GPUstats()
{
GASS.CUDA.CUDA cuda = new GASS.CUDA.CUDA(0, true);
Console.WriteLine("Device \"{0}\"", cuda.CurrentDevice.Name);
Console.WriteLine("\tCUDA Capability Major revision number:\t\t{0}", cuda.CurrentDevice.ComputeCapability.Major);
Console.WriteLine("\tCUDA Capability Minor revision number:\t\t{0}", cuda.CurrentDevice.ComputeCapability.Minor);
Console.WriteLine("\tTotal amount of global memory:\t\t\t{0} bytes", cuda.CurrentDevice.TotalMemory);
Console.WriteLine("\tNumber of multiprocessors:\t\t\t{0}", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.MultiProcessorCount));
Console.WriteLine("\tNumber of cores:\t\t\t\t{0}\n", 8 * cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.MultiProcessorCount));
Console.WriteLine("\tTotal amount of constant memory:\t\t{0}", cuda.CurrentDevice.Properties.TotalConstantMemory);
Console.WriteLine("\tTotal amount of shared memory per block:\t{0}", cuda.CurrentDevice.Properties.SharedMemoryPerBlock);
Console.WriteLine("\tTotal number of registers available per block:\t{0}", cuda.CurrentDevice.Properties.RegistersPerBlock);
Console.WriteLine("\tWarp size:\t\t\t\t\t{0}", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.WarpSize));
Console.WriteLine("\tMaximum number of threads per block:\t\t{0}", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.MaxThreadsPerBlock));
Console.WriteLine("\tMaximum sizes of each dimension of a block:\t{0} x {1} x {2}", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.MaxBlockDimX), cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.MaxBlockDimY), cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.MaxBlockDimZ));
Console.WriteLine("\tMaximum sizes of each dimension of a grid:\t{0} x {1} x {2}", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.MaxGridDimX), cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.MaxGridDimY), cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.MaxGridDimZ));
Console.WriteLine("\tMaximum memory pitch:\t\t\t\t{0}", cuda.CurrentDevice.Properties.MemoryPitch);
Console.WriteLine("\tTexture alignment:\t\t\t\t{0}", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.TextureAlignment));
Console.WriteLine("\tClock rate:\t\t\t\t\t{0} GHz\n", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.ClockRate) * 1e-6f);
Console.WriteLine("\tConcurrent copy and execution:\t\t\t{0}", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.ConcurrentKernels) == 1 ? "Yes" : "No");
Console.WriteLine("\tRun time limit on kernels:\t\t\t{0}", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.KernelExecTimeout) == 1 ? "Yes" : "No");
Console.WriteLine("\tIntegrated:\t\t\t\t\t{0}", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.Integrated) == 1 ? "Yes" : "No");
Console.WriteLine("\tSupport host page-locked memory mapping:\t{0}", cuda.GetDeviceAttribute(GASS.CUDA.CUDeviceAttribute.CanMapHostMemory) == 1 ? "Yes" : "No");
}
public override T Execute <T>(params object[] args)
{
if (string.IsNullOrEmpty(m_entryMethod))
{
throw new Exception("No method is loaded");
}
// Create cuda context
GASS.CUDA.CUDA cuda = new GASS.CUDA.CUDA(true);
// Load module (PTX) and get function
GASS.CUDA.Types.CUmodule module = cuda.LoadModule(m_ptx);
GASS.CUDA.Types.CUfunction func = cuda.GetModuleFunction(module, m_entryMethod);
// Create data
float[] A = new float[N];
float[] B = new float[N];
float[] C = new float[N];
Random rand = new Random();
for (int i = 0; i < N; i++)
{
A[i] = rand.Next(1000);
B[i] = rand.Next(1000);
C[i] = -1;
}
// Allocate data on device
GASS.CUDA.Types.CUdeviceptr dA = cuda.Allocate <float>(A);
GASS.CUDA.Types.CUdeviceptr dB = cuda.Allocate <float>(B);
GASS.CUDA.Types.CUdeviceptr dC = cuda.Allocate <float>(C);
// Copy data to device
cuda.CopyHostToDevice <float>(dA, A);
cuda.CopyHostToDevice <float>(dB, B);
cuda.CopyHostToDevice <float>(dC, C);
// Load parameters
int offset = 0;
cuda.SetParameter(func, offset, dA);
offset += Marshal.SizeOf(typeof(IntPtr));
cuda.SetParameter(func, offset, dB);
offset += Marshal.SizeOf(typeof(IntPtr));
cuda.SetParameter(func, offset, dC);
offset += Marshal.SizeOf(typeof(IntPtr));
cuda.SetParameterSize(func, (uint)offset);
// Setup execution
int blocksPerGrid = (N + threadsPerBlock - 1) / threadsPerBlock;
cuda.SetFunctionBlockShape(func, threadsPerBlock, 1, 1);
// Launch execution
cuda.Launch(func, blocksPerGrid, 1);
// Wait for and retrive result
cuda.SynchronizeContext();
cuda.CopyDeviceToHost <float>(dC, C);
// Clean device memory
cuda.Free(dA);
cuda.Free(dB);
cuda.Free(dC);
return(default(T));
}
