Get the latest tech news
Sorting algorithms with CUDA
A simple, whitespace theme for academics. Based on [*folio](https://github.com/bogoli/-folio) design.
I’ll take merge sort as our test algorithm because it nicely divides the problem into smaller subproblems with two equal halves, which is a good fit for parallel computing. Since CUDA does not efficiently handle recursion due to stack limitations, we implement an iterative approach for merge sort instead. Optimize further for performance on GPU by using shared memory, using thrust:sort at specific level in combination with my implementation like we do long multiplication in the karatsuba algorithm for size of n < 20 as I was taught in CSE 201 by Prof Sesh.
Or read this on Hacker News