问题
I am trying to generate set of indices below:
I have a cuda block that consists of 20 blocks(blockIdx:from 0 -19) with each block subdivided into 4 blocks (sub block Idx: 0,1,2 and 3).
I am trying to generate index pattern like this :
threadIdx (tid),SubBlockIdxA(SA),SubBlockIdxB(SB), BlockIdxA(BA),BlockIdxB(BB)
Required Obtained
tid SBA SBB BA BB SBA SBB BA BB
0 0 1 0 0 0 1 0 0
1 1 0 0 1 1 0 0 1
2 0 1 1 1 0 1 1 1
3 1 0 1 2 1 0 1 2
4 0 1 2 2 0 1 2 2
5 1 0 2 3 1 0 2 3
6 0 1 3 3 0 1 3 3
7 1 0 3 4 1 0 3 4
8 2 3 0 0 2 3 0 0
9 3 2 0 1 3 2 0 1
10 2 3 1 1 2 3 1 1
11 3 2 1 2 3 2 1 2
12 2 3 2 2 2 3 2 2
13 3 2 2 3 3 2 2 3
14 2 3 3 3 2 3 3 3
15 3 2 3 4 3 2 3 4
16 0 1 5 5 0 1 5 5
17 1 0 5 6 1 0 5 6
18 0 1 6 6 0 1 6 6
19 1 0 6 7 1 0 6 7
20 0 1 7 7 0 1 7 7
21 1 0 7 8 1 0 7 8
22 0 1 8 8 0 1 8 8
23 1 0 8 9 1 0 8 9
24 0 1 10 10 2 3 5 5
25 1 0 10 11 3 2 5 6
26 0 1 11 11 2 3 6 6
27 1 0 11 12 3 2 6 7
28 0 1 12 12 2 3 7 7
29 1 0 12 13 3 2 7 8
30 0 1 13 13 2 3 8 8
31 1 0 13 14 3 2 8 9
32 2 3 10 10 0 1 10 10
33 3 2 10 11 1 0 10 11
34 2 3 11 11 0 1 11 11
35 3 2 11 12 1 0 11 12
36 2 3 12 12 0 1 12 12
37 3 2 12 13 1 0 12 13
38 2 3 13 13 0 1 13 13
39 3 2 13 14 1 0 13 14
40 0 1 15 15 2 3 10 10
41 1 0 15 16 3 2 10 11
42 0 1 16 16 2 3 11 11
43 1 0 16 17 3 2 11 12
44 0 1 17 17 2 3 12 12
45 1 0 17 18 3 2 12 13
46 0 1 18 18 2 3 13 13
47 1 0 18 19 3 2 13 14
Please see my code below:
static __device__ void function()
{
uint16 uBlockIdxA, uBlockIdxB, uSubBlockIdxA, usubBlockIdxB;
if threadIdx.x < 48)
{
uint16 uY = threadIdx.x / 8;
uint16 uX = threadIdx.x - (uY * 8);
uSubBlockIdxA = ((uY & 0x01) << 1) + (uX & 0x01);
uSubBlockIdxB = ((uY & 0x01) << 1) + ((uX + 1) & 0x01);
uBlockIdxB = (uY >> 1) * 5 + ((1 + uX) >> 1);
uBlockIdxA = (uY >> 1) * 5 + ((0 + uX) >> 1);
func (uBlockIdxA, uBlockIdxB, uSubBlockIdxA, uSubBlockIdxB);
}
}
I am trying to think the logic to achieve what i am looking for. I am not right but not sure what i am missing.
logic as to how to generate this will be helpful. Code is appreciated. Please help.
Thanks in advance.
回答1:
code to generate the indices is below:
Split the threads into half since the indices from first half can be used for the second half with an offset of 10 and minor tweaks.
static __device__ void function()
{
uint16 uBlockIdxA, uBlockIdxB, uSubBlockIdxA, usubBlockIdxB;
if(threadIdx.x < 48)
{
uint16 uY = threadIdx.x / 8;
uint16 uX = threadIdx.x - (uY * 8);
if (threadIdx.x < 24)
{
uSubBlockIdxA = ((uY & 0x01) << 1) + (uX & 0x01);
uSubBlockIdxB = ((uY & 0x01) << 1) + ((uX + 1) & 0x01);
uBlockIdxB = (uY >> 1) * 5 + ((1 + uX) >> 1);
uBlockIdxA = (uY >> 1) * 5 + ((0 + uX) >> 1);
}
else if
{
uSubBlockIdxA = (((uY - 3) & 0x01) << 1) + (uX & 0x01);
uSubBlockIdxB = (((uY - 3) & 0x01) << 1) + ((uX + 1) & 0x01);
uBlockIdxB = ((uY - 3) >> 1) * 5 + ((1 + uX) >> 1) + 10;
uBlockIdxA = ((uY - 3) >> 1) * 5 + ((0 + uX) >> 1) + 10;
}
}
func (uBlockIdxA, uBlockIdxB, uSubBlockIdxA, uSubBlockIdxB);
}
来源：https://stackoverflow.com/questions/17334625/generate-index-using-cuda-c