AM3358: Slow execution of interpolation function

Part Number: AM3358

Hi, 
I am using Pocket Beagle bone with AM3358 processor. I want to perform an Interpolation function. When checked, the time taken to execute the function by the AM3358 processor on DebianOS was seen to be around 15-20 milli Sec. So, I decided to test the same code with a Linux system(Laptop), whose clock frequency is 2.0G Hz with an intel i3 processor. The time taken on Linux system was approx. 0-2 milli Sec. I assume the behavior of the AM3558 execution would be much better than the current result. I am using GCC for the compilation of the code. I am using the below command for the compilation with some optimization on AM3358.

 gcc speed_test.c -Ofast -mtune=cortex-a8 -mfpu=neon -funsafe-math-optimizations --param l1-cache-size=32 --param l2-cache-size=256

For Linux system(Laptop), I am using the below command:

 gcc speed_test.c -Ofast -funsafe-math-optimizations --param l1-cache-size=32 --param l2-cache-size=256

I think, I am doing something wrong in setting the configuration of GCC or the target processor. I have attached the code in the post.
Any help regarding the same would be appreciated.

Thanks & Regards,

Matt Dave.

/*Code for execution of Interpolation function*/
#include "stdio.h"
#include "float.h"
#include "sys/time.h"

#define PER_16  16
#define PHI_90	90

float dist;
int idx;
int i1;
float newDist;
/**
  * @brief  Find nearest point.
  * @param  value : to be found.
  * 		x : source pointer.
  * 		len : number of points.
  * @retval None
  */
int findNearestNeighbourIndex(float value,float *x,int len)
{
	idx=-1;
	dist=DBL_MAX;
	for(i1=0;i1<len;i1++)
	{
		newDist=(value-x[i1]);
		if((newDist>=0) && (newDist<dist))
		{
			dist=newDist;
			idx=i1;
		}
	}
return idx;
}
/*Function to get the current time*/
static long TimeM()
{
	struct timeval time;
	gettimeofday(&time,NULL);
	return time.tv_sec*1000 + time.tv_usec/1000;
}

/**
  * @brief  Find nearest point.
  * @param  value : to be found.
  * 		x : source pointer.
  * 		len : number of points.
  * 		indice : previous index.
  * @retval idx : index of the nearest point.
  */
int my_findNearestNeighbourIndex(float value,float *x,int len, int indice)
{
	volatile int s_index = 0, e_index = 0, index_offset;
	idx=-1;
	dist=DBL_MAX;

	index_offset = (PHI_90/3);
	
	s_index = (indice-5);//starting index
	if(s_index <= 0)//checks if index is out of minimum boundary limit
		s_index = 0;

	e_index = (indice+index_offset);//ending index
	if(e_index >= len)//checks if index is out of maximum boundary limit
		e_index = len;

	for(i1=(s_index);i1<(e_index);i1++)
	{
		newDist=(value-x[i1]);
		if((newDist>=0) && (newDist<dist))
		{
			dist=newDist;
			idx=i1;
		}
	}
return idx;
}

/**
  * @brief  Interpolation Function
  * @param  x : source pointer.
  * 		x_tam : length of x.
  * 		y : source pointer.
  * 		xx : pointer to the reference.
  * 		xx_tam : length of xx.
  * 		yy : destination pointer.
  * @retval None
  */
void interp1(float *x,int x_tam, float *y,float *xx,int xx_tam,float *yy)
{
	float dx,dy;

	float slope1, intercept1;
	register int i,indiceEnVector;


	indiceEnVector=findNearestNeighbourIndex(xx[0],x,x_tam);//find the nearest signal point index with respect to timedomain points
	for(i=0;i<=xx_tam;i++)
	{

		if(i>0)
		{

			indiceEnVector = my_findNearestNeighbourIndex(xx[i],x,x_tam,indiceEnVector);//find the nearest signal point index with respect to timedomain points

		}


		if(indiceEnVector != -1)//checks if nearest point is found
		{


			dx = (x[(indiceEnVector+1)] - x[indiceEnVector]); //change in x units
			dy= (y[(indiceEnVector+1)] - y[indiceEnVector]); //change in y units
			slope1 = (dy/dx); // slope
			intercept1 = (y[indiceEnVector]-x[indiceEnVector]*slope1); // intercept is calculated c=y-mx where m is slope and x,y are points with respect to index

			yy[i]=(slope1*xx[i]+intercept1); // straight line equation y=mx+c to find angle domain point.
		}
		else
		{

			yy[i]=DBL_MAX; // if no nearest point is found then index is -1 and default value is returned
		}

	}


}
int main()
{
	
    unsigned int n = 32768;
	float Block_Time[n];
	unsigned int nn = ((PER_16 * PHI_90)); 
	float angledomain[nn];
	float signal[n];
	float timedomain[nn];
	
	long start = TimeM();
	interp1(Block_Time,n,signal,timedomain,nn,angledomain);
	long end = TimeM();
	printf("Total Time : %ld\r\n",(end-start));    //Print the time taken

	return 0;
}