how to parallel this simple algorithm? someone has met this before?

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi,guys.

Now, I am confused by one ‘simple’ algorithm. It is like this:

  one array: 

index 0   1   2   3   4   5   6   7  8  9   

     a[]   0   0   1   0   0   1   0   1  0  1

I want to put the elements which has value!=0 in a into a new array, the elements in the array are the indexes indicate that which value is not 0.

   result: 

index   0   1   2   3

      result[]   2   5   7   9

Some guys has done this or met this before?

2

It is called “stream compaction” and it is implemented in CUDPP and Thrust.
There are plenty of papers describing it, just do a Google search

3

Thank you very much, mfatica.

I will google it. External Image