CUDA编程笔记(17)——Matrix transpose (shared memory)

An Efficient Matrix Transpose in CUDA C/C++Coalesced Transpose Via Shared Memory一节讲述如何使用shared memory高效地实现matrix transpose

__global__ void transposeCoalesced(float *odata, const float *idata)
{
  __shared__ float tile[TILE_DIM][TILE_DIM];

  int x = blockIdx.x * TILE_DIM + threadIdx.x;
  int y = blockIdx.y * TILE_DIM + threadIdx.y;
  int width = gridDim.x * TILE_DIM;

  for (int j = 0; j < TILE_DIM; j += BLOCK_ROWS)
 tile[threadIdx.y+j][threadIdx.x] = idata[(y+j)*width + x];

  __syncthreads();

  x = blockIdx.y * TILE_DIM + threadIdx.x;  // transpose block offset
  y = blockIdx.x * TILE_DIM + threadIdx.y;

  for (int j = 0; j < TILE_DIM; j += BLOCK_ROWS)
 odata[(y+j)*width + x] = tile[threadIdx.x][threadIdx.y + j];
}

(1)idataodata分别是表示1024X1024float元素的matrix的连续内存:

IMG_20170216_145959[1]

(2)关于blockIdxthreadIdx的取值,参考下面的图:

IMG_20170216_151045[1]

shared memory请参考下面的图:

IMG_20170216_151058[1]

(3)在下列代码中:

for (int j = 0; j < TILE_DIM; j += BLOCK_ROWS)
    tile[threadIdx.y+j][threadIdx.x] = idata[(y+j)*width + x];

每一个block32X8大小,需要循环4次,把一个block内容copytile这个shared memory中。idata是按行读取的,因此是coalesced

IMG_20170216_152636[1]

(4)最难理解的在最后一部分:

x = blockIdx.y * TILE_DIM + threadIdx.x;  // transpose block offset
y = blockIdx.x * TILE_DIM + threadIdx.y;

for (int j = 0; j < TILE_DIM; j += BLOCK_ROWS)
    odata[(y+j)*width + x] = tile[threadIdx.x][threadIdx.y + j];

对比从idata读取数据和写数据到odata

......
tile[threadIdx.y+j][threadIdx.x] = idata[(y+j)*width + x];
......
odata[(y+j)*width + x] = tile[threadIdx.x][threadIdx.y + j];
......

可以看到是把tile做了transpose的数据(行变列,列变行)传给odata。而确定需要把tile放到哪里位置的代码:

x = blockIdx.y * TILE_DIM + threadIdx.x;  // transpose block offset
y = blockIdx.x * TILE_DIM + threadIdx.y;

假设blockIdx.x31blockIdx.y0threadIdx.x1threadIdx.y2。根据上述代码,计算xy

x = 0 * 32 + 1;
y = 31 * 32 + 2;

根据下面的图,可以看到是把东北角的内容copy的西南角:

IMG_20170216_155616[1]

《CUDA编程笔记(17)——Matrix transpose (shared memory)》有2个想法

  1. 任意长宽的矩阵怎么处理? 这个地方是处理方阵, 感觉如果长宽不定或者不能整初的情况会比较麻烦….

发表评论

邮箱地址不会被公开。 必填项已用*标注

This site uses Akismet to reduce spam. Learn how your comment data is processed.