multi gpu - Atomic Exchange Sorting Algorithm in MultiGPU -
How can the nuclear exchange sorting algorithm be implemented in multi-GPU? Is there any reference available?
If you point out an algorithm that can be used, then this question will help A guideline to help answer
So, I pulled an algorithm from the following:
Here is the basic algorithm:
int main (zero) {Int array [5] ; // an array of integers int length = 5; // The string of the array. Int i, j; Int temporary; For some input {cout & lt; //} for (i = 0; i & lt; 5; i ++) & Lt; "Enter numbers:"; Cin & gt; & Gt; Array [i]; } // algorithm for (i = 0; i You may want to see this page for some source code that can help:
But if you Using OpenCL and the equation given above, you might want to do something like this:
Open a connection for each card.
Then, where there are external loops, send them to each, perhaps round robin for each card.
You will need a final type on a GPU to finish, but you would like to use a different algorithm, because this algorithm is best at single-threaded CPU.
Comments
Post a Comment