Noob: particle system

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi,

I’m noob with CUDA, but I try to create my own particle simulation, and I have problem with putting particles in right order. I want put particles in cube (they should look like cube of particles), usually i use this code (I have 8k particles, 3D):

int aa= 0;

	for(int xx = -10; xx <10; xx +=1 )

		for(int yy = -10; yy < 10; yy +=1)

			for(int zz = -10; zz < 10; zz +=1)

			{	

				particle[aa].r = Vector(xx, yy, zz);

				aa++;

			}

Now my .cu file:

__device__ float3 tab[8000];

__global__ void kernel(int n)

{

	int idx = blockIdx.x * blockDim.x + threadIdx.x;   //(0, 20)

	int aa = 0;

	if(idx < n)

	{

		for(int i = 0; i < 20; i++)

			for(int j = 0; j < 20; j++)

				tab[aa++].r = make_float3(idx, i, j);

	}

}

Can someone help me, or explain me what I’m doing wrong.

Sry for my English and thx for help.

2

You are only accessing the first 400 indices of your tab array. Changing tab[aa++] to tab[400*idx+aa++] should correct this.

3

Thx for your response, but still it doesn’t work. Now I can see 1 k particles:/

Any other sugestions?

4

Yes, instead of launching 20 threads, launch 8000 threads and let each one fill an index in the array. Like:

z = idx/400

y = (idx - z*400)/20

x = idx - z400 - y20

tab[idx] = make_float3(x,y,z)