code example
OpenMP multi-threading
Cuda nVidia Computer Unified Device Architecture
GPU computing

#include <omp.h>
#include <cuda.h>

float e;

// kernel
__global__ void sub1(float* fx, float* fy, float* fe) {
#define BLOCK (512)
  int t = threadIdx.x; // builtin
  int b = blockIdx.x; // builtin
  float e;
  __shared__ float se[BLOCK];
  __shared__ float sx[BLOCK];
  __shared__ float sy[BLOCK+2];
  // copy from device to processor memory
  sx[t] = fx[BLOCK*b+t];
  sy[t] = fy[BLOCK*b+t];
  if (t<2)
     sy[t+BLOCK] = fy[BLOCK*b+t+BLOCK];
  __syncthreads();

  // do computation
  sx[t] += ( sy[t+2] + sy[t] )*.5;
  e = sy[t+1] * sy[t+1];
  // copy to device memory
  fx[BLOCK*b+t] = sx[t];
  // reduction
  se[t] = e;
  __syncthreads();
  if (t<256) {
     se[t] += se[t+256];
     __syncthreads();
  }
  if (t<128) {
     se[t] += se[t+128];
     __syncthreads();
  }
  if (t<64) {
     se[t] += se[t+64];
     __syncthreads();
  }
  if (t<32) { // warp size
     se[t] += se[t+32];
     se[t] += se[t+16];
     se[t] += se[t+8];
     se[t] += se[t+4];
     se[t] += se[t+2];
     se[t] += se[t+1];
  }
  if (t==0)
     fe[b] = se[0];
}

int main(int argc, char *argv[]) {
  int n = ...;
  float *x, *y;
  x = new float[n+1];
  y = new float[n+1];

  ... // fill x, y

  e = 0;
  #pragma omp parallel
  {
  int p = omp_get_thread_num();
  int num = omp_get_num_threads();
  // pick GPU
  cudaSetDevice(p);
  // allocate GPU memory
  float *fx, *fy, *fe;
  cudaMalloc((void**)&fx, (n-1+2) * sizeof(float));
  cudaMalloc((void**)&fy, (n-1+2) * sizeof(float));
  cudaMalloc((void**)&fe, (n-1+2)/BLOCK * sizeof(float));
  float *de = new float[(n-1+2)/BLOCK];
  // copy to GPU memory
  cudaMemcpy(fx+1, &x[1],
   (n-1) * sizeof(float), cudaMemcpyHostToDevice);
  cudaMemcpy(fy, &y[1-1],
   (n-1+2) * sizeof(float), cudaMemcpyHostToDevice);
  dim3 dimBlock(BLOCK, 1, 1);
  dim3 dimGrid((n-1+2)/BLOCK, 1, 1);

  int n0 = 1+((n-1)*p)/num;
  int n1 = 1+((n-1)*(p+1))/num;
  // call GPU
  sub1<<<dimGrid, dimBlock>>>(fx, fy, fe);
  // copy to host memory
  cudaMemcpy(fx+1, &x[n0], (n1-n0) * sizeof(float),
   cudaMemcpyDeviceToHost);
  cudaMemcpy(fe, &de[n0-1], (n1-n0+2)/BLOCK * sizeof(float),
   cudaMemcpyDeviceToHost);
  // release GPU memory
  cudaFree(fe);
  cudaFree(fy);
  cudaFree(fx);
  float e_local = 0;
  for (int i=0; i<(n1-n0+2)/BLOCK; ++i)
   e_local += de[i];
  #pragma omp atomic
  e += e_local;
  delete[] de;
  }

  ... // output x, e

  delete[] x, y;
  return 0;
}

[start] [references] [download] [install]