reduce2.cu File Reference

Reduces M x N matrix. Returns column results for a generic reduction operation. More...

#include "../include/mycuda.h"
#include "../include/mycuda_reduce.h"

Go to the source code of this file.

Functions
template<typename T , typename R >
__global__ void	reduce2 (T *xsum, T *x, int n, int stride, R op)
	Reduction kernel. More...
int	main ()

Detailed Description

Reduces M x N matrix. Returns column results for a generic reduction operation.

Notes

• Columns can be of arbitrary length, but one block reduces each column.

Definition in file reduce2.cu.

Function Documentation

int main

(

void

)

Definition at line 70 of file reduce2.cu.

template<typename T , typename R >

__global__ void reduce2	(	T *	xsum,
		T *	x,
		int	n,
		int	stride,
		R	op
	)

Reduction kernel.

Notes

• Uses dynamic allocation for shared memory. Allocation size is determined by configuration of kernel launch.
• Each block reduces one column.

Requires

• blocksize >= 32
• shared_memory_size = 2*blocksize*sizeof(T)
• gridsize = ncolumns of matrix

Definition at line 30 of file reduce2.cu.