cuda数组内存分配

来源：互联网发布：全程软件测试朱少民编辑：程序博客网时间：2024/05/18 01:27

http://www.cnblogs.com/traceorigin/archive/2013/04/12/3016540.html

cuda内存分配真是乱啊，这次总结一下数组的分配。

概述：数组分配可以通过cudaMallocArray()和cudaMalloc3DArray()

1、cudaMallocArray()

cudaError_t cudaMallocArray    (    struct cudaArray **     array,    const struct cudaChannelFormatDesc *     desc,    size_t     width,    size_t     height = 0,    unsigned int     flags = 0     )

例：2DArray分配

1 cudaArray *d_volumeArray;2 cudaChannelFormatDesc desc = cudaCreateChannelDesc(32, 0, 0, 0, cudaChannelFormatKindFloat);3 cudaMallocArray(&d_volumeArray, &desc, 128, 128);4 cudaMemcpyToArray(d_volumeArray, 0, 0, h_volume, sizeof(float)*128*128, cudaMemcpyDeviceToDevice);

2、cudaMalloc3DArray()

1 cudaError_t cudaMalloc3DArray    (    2     struct cudaArray **     array,3     const struct cudaChannelFormatDesc *     desc,4     struct cudaExtent     extent,5     unsigned int     flags = 0     6 )

例：

1 cudaChannelFormatDesc channelDesc = cudaCreateChannelDesc<uchar>();2 cudaArray *d_volumeArray;3 cudaMalloc3DArray(&d_volumeArray, &channelDesc, volumSize);