我在检索要托管的设备数据时遇到了问题。在下面的示例程序中。我在主机中创建了一个hostData数组,并为其提供了值。然后,我将hostData传递给deviceData变量,并再次将相同的数据传递给新的主机数据数组,即。hostDataFinal。当我打印存储在hostDataFinal中的值时,最多只有几个值,输出是ok,rest显示为垃圾值。所有值都应该与hostData数组中的值相同。请帮我一把。
#include "cuda_runtime.h"
#include "device_launch_parameters.h"
#include <iostream>
int main()
{
int totalData = 20;
int *hostData;
hostData = new int[totalData];
std::cout<<std::endl<<std::endl<<"HostData Values : "<<std::endl;
for (int i=0; i<totalData ;i++)
{
hostData[i]= i+1;
std::cout<<hostData[i]<<" ";
}
int *deviceData;
int *hostDataFinal;
cudaMalloc((void**)&deviceData,totalData*sizeof(int));
hostDataFinal = new int[totalData];
cudaMemcpy(deviceData,hostData,totalData,cudaMemcpyHostToDevice);
cudaMemcpy(hostDataFinal,deviceData,totalData,cudaMemcpyDeviceToHost);
std::cout<<std::endl<<std::endl<<"HostDataFinal After Device Transfer Values : "<<std::endl;
for (int i=0; i<totalData ;i++)
{
std::cout<<hostDataFinal[i]<<" ";
}
free(hostData);
free(hostDataFinal);
cudaFree(deviceData);
return 0;
}
该方案的产出:
HostData值:1 2 3 4 5 6 7 9 10 11 13 14 15 16 17 18 19 20
设备传输值后的HostDataFinal:
1 2 3 4 5 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451 -842150451-842150451-842150451
实际产出应是:
HostData值:1 2 3 4 5 6 7 9 10 11 13 14 15 16 17 18 19 20
设备传输值后的HostDataFinal:
1 2 3 4 5 6 7 8 10 11 12 14 15 16 17 18 19 20
谢谢你的帮助。
发布于 2013-12-05 18:04:33
你的问题就在这些线上。cudaMemcpy
中的第三个参数是要复制的字节数。
cudaMemcpy(deviceData,hostData,totalData,cudaMemcpyHostToDevice);
cudaMemcpy(hostDataFinal,deviceData,totalData,cudaMemcpyDeviceToHost);
应该是
cudaMemcpy(deviceData,hostData,totalData *sizeof(int),cudaMemcpyHostToDevice);
cudaMemcpy(hostDataFinal,deviceData,totalData*sizeof(int),cudaMemcpyDeviceToHost);
还要确保处理返回错误代码,如here所示
https://stackoverflow.com/questions/20414505
复制相似问题