improved RBGrid->getPosition by eschewing modulo; now performance on par with 3D kernel

ComputeGridGrid.cuh

@@ -34,7 +34,7 @@ void computeGridGridForce(const RigidBodyGrid* rho, const RigidBodyGrid* u,
 	/* Vector3 tmpf  = Vector3(0.0f); */
 	/* float tmpe = 0.0f; */
 	/* const ForceEnergy fe = ForceEnergy( tmpf, tmpe); */
-	const ForceEnergy fe = u->interpolateForceD( u_ijk_float ); /* in coord frame of u */
+	const ForceEnergy fe = u->interpolateForceDLinearly( u_ijk_float ); /* in coord frame of u */
 	force[tid] = fe.f;
 
 	const float r_val = rho->val[r_id]; /* maybe move to beginning of function?  */

RigidBodyGrid.cu

@@ -178,11 +178,17 @@ float RigidBodyGrid::getValue(int ix, int iy, int iz) const {
 }
 
 Vector3 RigidBodyGrid::getPosition(const int j) const {
-	const int iz = j%nz;
-	const int iy = (j/nz)%ny;
-	const int ix = j/(nz*ny);
+	/* const int iz = j%nz; */
+	/* const int iy = (j/nz)%ny; */
+	/* const int ix = j/(nz*ny); */
+	const int jy = j/nz;
+	const int jx = jy/ny;
 
-	return Vector3(ix,iy,iz);
+	const int iz = j - jy*nz;
+	const int iy = jy - jx*ny;
+	// const int ix = jx;
+
+	return Vector3(jx,iy,iz);
 }
 
 Vector3 RigidBodyGrid::getPosition(int j, Matrix3 basis, Vector3 origin) const {