Sign in to follow this  
  • entries
    455
  • comments
    639
  • views
    422432

TLM Optimisation...

Sign in to follow this  
_the_phantom_

85 views

I was walking to the pub last night, letting my mind wander over the code I'm going to have to do when a thought struck me; What I'd planned todo in 3 passes I could do in one and with only two render targets.

The code, as mentioned before, peforms a sum of all the energy from different directions into the current point, then in a seperate loop works out how much energy a given cell has based on what is being sent into it from surrounding cells, it then calculates the position.


// Inner loops
for(int x = 0; x < width; ++x)
{
for(int y = 0; y < height; ++y)
{
Sn[x][y] = 0.5 * (-In[x][y] + Is[x][y] + Ie[x][y] + Iw[x][y]);
Sw[x][y] = 0.5 * ( In[x][y] - Is[x][y] + Ie[x][y] + Iw[x][y]);
Se[x][y] = 0.5 * ( In[x][y] + Is[x][y] - Ie[x][y] + Iw[x][y]);
Sw[x][y] = 0.5 * ( In[x][y] + Is[x][y] + Ie[x][y] - Iw[x][y]);
}
}
for(int x = 0; x < width; ++x)
{
for(int y = 0; y < height; ++y)
{
Ie[x][y] = Sw[x+1][y];
Iw[x][y] = Se[x-1][y];
In[x][y] = Sw[x][y-1];
Is[x][y] = Se[x][y+1];

nodevalue[x][y] = (/* maths involving the above sink array elements*/);
}
}





It occured to me that the first two steps could be combined into one and the final value worked out from that. Ping-ponging two render buffers means I can read from one and write to another with no problems. This effectly turns an O(n^2) op into an O(n) operation...

I'm half wondering how such a simple optimisation was missed? That said, the text describes API as 'advanced programming interface', which makes me cry...
Sign in to follow this  


1 Comment


Recommended Comments

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now