/*

 * Copyright (C) 2003 Texas Instruments Incorporated

 * All Rights Reserved

 */

/*

 *---------dotpV2_timer2.c---------

 * This file contains 'dot product version 2' routine, which is raw C

 * with some optimization by giving more information to compiler

 * in order to unroll the loop and reduce the cycle count.

 */



int dotprod_ver2 (const short *restrict x, const short *restrict h, int nx)

{

	int i, sum = 0;

   

	//Assumption: Vectors x and h are double-word aligned

	_nassert((int)x % 8 == 0);

	_nassert((int)h % 8 == 0);

   

    //Assumption: Counter nx (vector length) is atleast 36 and multiple of 4

	#pragma MUST_ITERATE(36,,4)

	for (i=0; i<nx; i++)

		sum += x[i] * h[i];



	return sum;

}