//
// Author: Mengjuei Hsieh, UC Irvine, Irvine, CA
//
//This wrapper is inspired by Ian Ollmann's sample code for Apple Computer
//To compile:
//	gcc -c -O3 -fast -funroll-loops -faltivec vector.c
//

#include <vecLib/vfp.h>

void vrsqrtd_wrapper( double *in, double *out, int *n )
{
  int i;
  int loopCount = *n / vec_step( vector float );
        union
        {
	  float scalarArray[ *n + 3 ];
	  vector float v[ loopCount + 1 ];
        }buffer;

        for( i = 0; i < *n ; i++ )
	  buffer.scalarArray[i] = (float) in[i];
        buffer.scalarArray[*n+1] = 1;
        buffer.scalarArray[*n+2] = 1;
        buffer.scalarArray[*n+3] = 1;

        for( i = 0; i < loopCount + 1; i++ )
	  buffer.v[i] = vrsqrtf( buffer.v[i] );

        for( i = 0; i < *n; i++ )
	  out[i] = (double) buffer.scalarArray[i];
}

/*
void vrsqrtf_wrapper( float *in, float *out, int *n )
{
  int i;
  int loopCount = *n / vec_step( vector float );
        union
        {
	  float scalarArray[ *n + 3 ];
	  vector float v[ loopCount + 1 ];
        }buffer;

        for( i = 0; i < *n ; i++ )
	  buffer.scalarArray[i] = in[i];
        buffer.scalarArray[*n+1] = 1;
        buffer.scalarArray[*n+2] = 1;
        buffer.scalarArray[*n+3] = 1;

        for( i = 0; i < loopCount + 1; i++ )
	  buffer.v[i] = vrsqrtf( buffer.v[i] );

        for( i = 0; i < *n; i++ )
	  out[i] = buffer.scalarArray[i];
}
*/