相关疑难解决方法(0)

我正在测量标准和1Dtexture访问内存之间的差异.为此,我创建了两个内核

__global__ void texture1D(float* doarray,int size)
{
  int index;
  //calculate each thread global index
  index=blockIdx.x*blockDim.x+threadIdx.x;
  //fetch global memory through texture reference
  doarray[index]=tex1Dfetch(texreference,index);
  return;
}
__global__ void standard1D(float* diarray, float* doarray, int size)
{
  int index;
  //calculate each thread global index
  index=blockIdx.x*blockDim.x+threadIdx.x;
  //fetch global memory through texture reference
  doarray[index]= diarray[index];
  return;
}

然后,我调用eache内核来测量它所花费的时间:

//copy array from host to device memory
  cudaMemcpy(diarray,harray,sizeof(float)*size,cudaMemcpyHostToDevice);

  checkCuda( cudaEventCreate(&startEvent) );
  checkCuda( cudaEventCreate(&stopEvent) );
  checkCuda( cudaEventRecord(startEvent, 0) );

  //bind texture reference with linear memory
  cudaBindTexture(0,texreference,diarray,sizeof(float)*size);

  //execute device kernel …