Sorting pairs of numbers using AMP
I need numbers next to each other to be sorted (in increasing order).
On the CPU I'd do something like this:
1 2 3 4 5
|
for(int i=0; i < length; i+=2) {
if(a[i]>a[i+1]) {
switch places...
}
}
|
But how would I do this using parallel_for_each (C++AMP) ?
I need this for some algorithm that works with very long arrays and I think GPU would do this faster than CPU (even if I use all threads).
Try websearch with "GPGPU sort".
My question is: is there a way to execute the kernel for every Second element of an array?
Of course. Lets say that you have N threads. Each has index from [0..N[
You have array with at least 2*N elements.
Let kernel compute: int arrayIndex = 2*threadIndex;
Topic archived. No new replies allowed.