float
ComputeGravitation_SIMD_openmp(
float *force[3],
float *pos[4],
float *mass,
float softeningSquared,
size_t N
)
{
chTimerTimestamp start, end;
chTimerGetTime( &start );
#pragma omp parallel for
for (int i = 0; i < N; i++)
{
v4sf ax = vec_zero;
v4sf ay = vec_zero;
v4sf az = vec_zero;
v4sf *px = (v4sf *) pos[0];
v4sf *py = (v4sf *) pos[1];
v4sf *pz = (v4sf *) pos[2];
v4sf *pmass = (v4sf *) mass;
v4sf x0 = _vec_set_ps1( pos[0][i] );
v4sf y0 = _vec_set_ps1( pos[1][i] );
v4sf z0 = _vec_set_ps1( pos[2][i] );
for ( int j = 0; j < N/4; j++ ) {
bodyBodyInteraction(
ax, ay, az,
x0, y0, z0,
px[j], py[j], pz[j], pmass[j],
_vec_set_ps1( softeningSquared ) );
}
// Accumulate sum of four floats in the AltiVec register
force[0][i] = _vec_sum( ax );
force[1][i] = _vec_sum( ay );
force[2][i] = _vec_sum( az );
}
chTimerGetTime( &end );
return (float) chTimerElapsedTime( &start, &end ) * 1000.0f;
}
//
// Get elapsed Time
//
double
getTime() { return chTimerElapsedTime( &m_start, &m_end ); };
