void revert_kernel_c_(int *xmin,int *xmax,int *ymin,int *ymax,
double *density0,
double *density1,
double *energy0,
double *energy1)
{
int x_min=*xmin;
int x_max=*xmax;
int y_min=*ymin;
int y_max=*ymax;
int j,k;
{
for (k=y_min;k<=y_max;k++) {
#pragma ivdep
for (j=x_min;j<=x_max;j++) {
density1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]=density0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
}
}
for (k=y_min;k<=y_max;k++) {
#pragma ivdep
for (j=x_min;j<=x_max;j++) {
energy1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]=energy0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
}
}
}
}
void ideal_gas_kernel_c_(int *xmin,int *xmax,int *ymin,int *ymax,
double *density,
double *energy,
double *pressure,
double *soundspeed)
{
int x_min=*xmin;
int x_max=*xmax;
int y_min=*ymin;
int y_max=*ymax;
int j,k;
double sound_speed_squared,v,pressurebyenergy,pressurebyvolume;
#pragma omp parallel private(j)
{
#pragma omp for private(v,pressurebyenergy,pressurebyvolume,sound_speed_squared)
for (k=y_min; k<=y_max; k++) {
#pragma ivdep
for (j=x_min; j<=x_max; j++) {
v=1.0/density[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
pressure[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]=(1.4-1.0)*density[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
*energy[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
pressurebyenergy=(1.4-1.0)*density[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
pressurebyvolume=-density[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]*pressure[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
sound_speed_squared=v*v*(pressure[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]*pressurebyenergy-pressurebyvolume);
soundspeed[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]=sqrt(sound_speed_squared);
}
}
}
}
void unpack_left_right_buffers_c_(int *xmin,int *xmax,int *ymin,int *ymax,
int *chnk_lft,int *chnk_rght,int *xtrnl_fc,
int *xinc,int *yinc,int *dpth,int *sz,
double *field, double *left_rcv_buffer, double *right_rcv_buffer)
{
int x_min=*xmin;
int x_max=*xmax;
int y_min=*ymin;
int y_max=*ymax;
int chunk_left=*chnk_lft;
int chunk_right=*chnk_rght;
int external_face=*xtrnl_fc;
int x_inc=*xinc;
int y_inc=*yinc;
int depth=*dpth;
int j,k,index;
{
if(chunk_left!=external_face) {
for (k=y_min-depth;k<=y_max+y_inc+depth;k++) {
#pragma ivdep
for (j=1;j<=depth;j++) {
index=j+(k+depth-1)*depth;
field[FTNREF2D(x_min-j,k,x_max+4+x_inc,x_min-2,y_min-2)]=left_rcv_buffer[FTNREF1D(index,1)];
}
}
}
if(chunk_right!=external_face) {
for (k=y_min-depth;k<=y_max+y_inc+depth;k++) {
#pragma ivdep
for (j=1;j<=depth;j++) {
index=j+(k+depth-1)*depth;
field[FTNREF2D(x_max+x_inc+j,k,x_max+4+x_inc,x_min-2,y_min-2)]=right_rcv_buffer[FTNREF1D(index,1)];
}
}
}
}
}
void reset_field_kernel_c_(int *xmin,int *xmax,int *ymin,int *ymax,
double *density0,
double *density1,
double *energy0,
double *energy1,
double *xvel0,
double *xvel1,
double *yvel0,
double *yvel1)
{
int x_min=*xmin;
int x_max=*xmax;
int y_min=*ymin;
int y_max=*ymax;
int j,k;
#pragma omp parallel
{
#pragma omp for private(j)
for (k=y_min;k<=y_max;k++) {
#pragma ivdep
for (j=x_min;j<=x_max;j++) {
density0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]=density1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
}
}
#pragma omp for private(j)
for (k=y_min;k<=y_max;k++) {
#pragma ivdep
for (j=x_min;j<=x_max;j++) {
energy0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]=energy1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
}
}
#pragma omp for private(j)
for (k=y_min;k<=y_max+1;k++) {
#pragma ivdep
for (j=x_min;j<=x_max+1;j++) {
xvel0[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=xvel1[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
}
}
#pragma omp for private(j)
for (k=y_min;k<=y_max+1;k++) {
#pragma ivdep
for (j=x_min;j<=x_max+1;j++) {
yvel0[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=yvel1[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
}
}
}
}
void unpack_top_bottom_buffers_c_(int *xmin,int *xmax,int *ymin,int *ymax,
int *chnk_bttm,int *chnk_tp,int *xtrnl_fc,
int *xinc,int *yinc,int *dpth,int *sz,
double *field, double *bottom_rcv_buffer, double *top_rcv_buffer)
{
int x_min=*xmin;
int x_max=*xmax;
int y_min=*ymin;
int y_max=*ymax;
int chunk_top=*chnk_tp;
int chunk_bottom=*chnk_bttm;
int external_face=*xtrnl_fc;
int x_inc=*xinc;
int y_inc=*yinc;
int depth=*dpth;
int j,k,index;
{
if(chunk_bottom!=external_face) {
for (k=1;k<=depth;k++) {
for (j=x_min-depth;j<=x_max+x_inc+depth;j++) {
index=j+depth+(k-1)*(x_max+x_inc+(2*depth));
field[FTNREF2D(j,y_min-k,x_max+4+x_inc,x_min-2,y_min-2)]=bottom_rcv_buffer[FTNREF1D(index,1)];
}
}
}
if(chunk_top!=external_face) {
for (k=1;k<=depth;k++) {
for (j=x_min-depth;j<=x_max+x_inc+depth;j++) {
index=j+depth+(k-1)*(x_max+x_inc+(2*depth));
field[FTNREF2D(j,y_max+y_inc+k,x_max+4+x_inc,x_min-2,y_min-2)]=top_rcv_buffer[FTNREF1D(index,1)];
}
}
}
}
}
void accelerate_kernel_c_(int *xmin,int *xmax,int *ymin,int *ymax,
double *dbyt,
double *xarea,
double *yarea,
double *volume,
double *density0,
double *pressure,
double *viscosity,
double *xvel0,
double *yvel0,
double *xvel1,
double *yvel1)
{
int x_min=*xmin;
int x_max=*xmax;
int y_min=*ymin;
int y_max=*ymax;
double dt=*dbyt;
int j,k,err;
double nodal_mass;
double stepby_mass_s;
for (k=y_min;k<=y_max+1;k++) {
#pragma ivdep
for (j=x_min;j<=x_max+1;j++) {
nodal_mass=(density0[FTNREF2D(j-1,k-1,x_max+4,x_min-2,y_min-2)]*volume[FTNREF2D(j-1,k-1,x_max+4,x_min-2,y_min-2)]
+density0[FTNREF2D(j ,k-1,x_max+4,x_min-2,y_min-2)]*volume[FTNREF2D(j ,k-1,x_max+4,x_min-2,y_min-2)]
+density0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]*volume[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
+density0[FTNREF2D(j-1,k ,x_max+4,x_min-2,y_min-2)]*volume[FTNREF2D(j-1,k ,x_max+4,x_min-2,y_min-2)])
*0.25;
stepby_mass_s=0.5*dt/nodal_mass;
xvel1[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=xvel0[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
-stepby_mass_s
*(xarea[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
*(pressure[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]-pressure[FTNREF2D(j-1,k ,x_max+4,x_min-2,y_min-2)])
+xarea[FTNREF2D(j ,k-1,x_max+5,x_min-2,y_min-2)]
*(pressure[FTNREF2D(j ,k-1,x_max+4,x_min-2,y_min-2)]-pressure[FTNREF2D(j-1,k-1,x_max+4,x_min-2,y_min-2)]));
yvel1[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=yvel0[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
-stepby_mass_s
*(yarea[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
*(pressure[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]-pressure[FTNREF2D(j ,k-1,x_max+4,x_min-2,y_min-2)])
+yarea[FTNREF2D(j-1,k ,x_max+4,x_min-2,y_min-2)]
*(pressure[FTNREF2D(j-1,k ,x_max+4,x_min-2,y_min-2)]-pressure[FTNREF2D(j-1,k-1,x_max+4,x_min-2,y_min-2)]));
xvel1[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=xvel1[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
-stepby_mass_s
*(xarea[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
*(viscosity[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]-viscosity[FTNREF2D(j-1,k ,x_max+4,x_min-2,y_min-2)])
+xarea[FTNREF2D(j ,k-1,x_max+5,x_min-2,y_min-2)]
*(viscosity[FTNREF2D(j ,k-1,x_max+4,x_min-2,y_min-2)]-viscosity[FTNREF2D(j-1,k-1,x_max+4,x_min-2,y_min-2)]));
yvel1[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=yvel1[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
-stepby_mass_s
*(yarea[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
*(viscosity[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]-viscosity[FTNREF2D(j ,k-1,x_max+4,x_min-2,y_min-2)])
+yarea[FTNREF2D(j-1,k ,x_max+4,x_min-2,y_min-2)]
*(viscosity[FTNREF2D(j-1,k ,x_max+4,x_min-2,y_min-2)]-viscosity[FTNREF2D(j-1,k-1,x_max+4,x_min-2,y_min-2)]));
}
}
}
void generate_chunk_kernel_c_(int *xmin,int *xmax,int *ymin,int *ymax,
double *vertexx,
double *vertexy,
double *cellx,
double *celly,
double *density0,
double *energy0,
double *xvel0,
double *yvel0,
int *nmbr_f_stts,
double *state_density,
double *state_energy,
double *state_xvel,
double *state_yvel,
double *state_xmin,
double *state_xmax,
double *state_ymin,
double *state_ymax,
double *state_radius,
int *state_geometry,
int *g_rct,
int *g_crc,
int *g_pnt)
{
int x_min=*xmin;
int x_max=*xmax;
int y_min=*ymin;
int y_max=*ymax;
int number_of_states=*nmbr_f_stts;
int g_rect=*g_rct;
int g_circ=*g_crc;
int g_point=*g_pnt;
START_PROFILING;
/* State 1 is always the background state */
#pragma omp parallel for
for (int k = y_min-2; k <= y_max+2; k++)
{
#pragma ivdep
for (int j = x_min-2; j <= x_max+2; j++)
{
energy0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)] =
state_energy[FTNREF1D(1,1)];
}
}
#pragma omp parallel for
for (int k=y_min-2;k<=y_max+2;k++)
{
#pragma ivdep
for (int j=x_min-2;j<=x_max+2;j++)
{
density0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)] =
state_density[FTNREF1D(1,1)];
}
}
#pragma omp parallel for
for (int k = y_min-2; k<= y_max+2; k++)
{
#pragma ivdep
for (int j = x_min-2; j <= x_max+2; j++)
{
xvel0[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)] =
state_xvel[FTNREF1D(1,1)];
}
}
#pragma omp parallel for
for (int k = y_min-2;k <= y_max+2; k++)
{
#pragma ivdep
for (int j = x_min-2; j <= x_max+2; j++)
{
yvel0[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)] =
state_yvel[FTNREF1D(1,1)];
}
}
for (int state = 2; state <= number_of_states; state++)
{
/* Could the velocity setting be thread unsafe? */
double x_cent = state_xmin[FTNREF1D(state,1)];
double y_cent = state_ymin[FTNREF1D(state,1)];
#pragma omp parallel for
for (int k = y_min-2; k <= y_max+2; k++)
{
#pragma ivdep
for (int j = x_min-2; j <= x_max+2; j++)
{
if(state_geometry[FTNREF1D(state,1)] == g_rect )
{
if(vertexx[FTNREF1D(j+1,x_min-2)]>=state_xmin[FTNREF1D(state,1)] &&
vertexx[FTNREF1D(j,x_min-2)]<state_xmax[FTNREF1D(state,1)])
{
if(vertexy[FTNREF1D(k+1,y_min-2)]>=state_ymin[FTNREF1D(state,1)] &&
vertexy[FTNREF1D(k,y_min-2)]<state_ymax[FTNREF1D(state,1)])
{
density0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)] =
state_density[FTNREF1D(state,1)];
energy0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)] =
state_energy[FTNREF1D(state,1)];
for (int kt = k; kt <= k+1; kt++)
{
for (int jt = j; jt <= j+1; jt++)
{
xvel0[FTNREF2D(jt,kt,x_max+5,x_min-2,y_min-2)] =
state_xvel[FTNREF1D(state,1)];
yvel0[FTNREF2D(jt,kt,x_max+5,x_min-2,y_min-2)] =
state_yvel[FTNREF1D(state,1)];
}
}
}
}
}
else if(state_geometry[FTNREF1D(state,1)] == g_circ)
{
double radius = sqrt(
(cellx[FTNREF1D(j,x_min-2)]-x_cent) *
(cellx[FTNREF1D(j,x_min-2)]-x_cent) +
(celly[FTNREF1D(k,y_min-2)]-y_cent) *
(celly[FTNREF1D(k,y_min-2)]-y_cent));
if(radius <= state_radius[FTNREF1D(state,1)])
{
density0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)] =
state_density[FTNREF1D(state,1)];
energy0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)] =
state_density[FTNREF1D(state,1)];
for (int kt = k; kt <= k+1; kt++)
{
for (int jt = j; jt <= j+1; jt++)
{
xvel0[FTNREF2D(jt,kt,x_max+5,x_min-2,y_min-2)] =
state_xvel[FTNREF1D(state,1)];
yvel0[FTNREF2D(jt,kt,x_max+5,x_min-2,y_min-2)] =
state_yvel[FTNREF1D(state,1)];
}
}
}
}
else if(state_geometry[FTNREF1D(state,1)] == g_point)
{
if(vertexx[FTNREF1D(j,x_min-2)] == x_cent &&
vertexy[FTNREF1D(j,x_min-2)]==y_cent)
{
density0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)] =
state_density[FTNREF1D(state,1)];
energy0[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)] =
state_density[FTNREF1D(state,1)];
for (int kt = k; kt <= k+1; kt++)
{
for (int jt = j; jt <= j+1; jt++)
{
xvel0[FTNREF2D(jt,kt,x_max+5,x_min-2,y_min-2)] =
state_xvel[FTNREF1D(state,1)];
yvel0[FTNREF2D(jt,kt,x_max+5,x_min-2,y_min-2)] =
state_yvel[FTNREF1D(state,1)];
}
}
}
}
}
}
}
STOP_PROFILING(__func__);
}
void calc_dt_kernel_c_(int *xmin,int *xmax,int *ymin,int *ymax,
double *gsmall,double *gbig,double *mindt,
double *dtcsafe,
double *dtusafe,
double *dtvsafe,
double *dtdivsafe,
double *xarea,
double *yarea,
double *cellx,
double *celly,
double *celldx,
double *celldy,
double *volume,
double *density0,
double *energy0,
double *pressure,
double *viscosity,
double *soundspeed,
double *xvel0,
double *yvel0,
double *dt_min,
double *dtminval,
int *dtlcontrol,
double *xlpos,
double *ylpos,
int *jldt,
int *kldt,
int *smll)
{
int x_min=*xmin;
int x_max=*xmax;
int y_min=*ymin;
int y_max=*ymax;
double g_small=*gsmall;
double g_big=*gbig;
double dt_min_val=*dtminval;
double dtc_safe=*dtcsafe;
double dtu_safe=*dtusafe;
double dtv_safe=*dtvsafe;
double dtdiv_safe=*dtdivsafe;
double min_dt=*mindt;
int dtl_control=*dtlcontrol;
double xl_pos=*xlpos;
double yl_pos=*ylpos;
int j_ldt=*jldt;
int k_ldt=*kldt;
int small=*smll;
int j,k;
double div,dsx,dsy,dtut,dtvt,dtct,dtdivt,cc,dv1,dv2,jk_control;
small=0;
dt_min_val = g_big;
jk_control=1.1;
#pragma omp parallel
{
#pragma omp for private(dsx,dsy,cc,dv1,dv2,div,dtct,dtut,dtvt,dtdivt,j,k)
for (k=y_min;k<=y_max;k++) {
#pragma ivdep
for (j=x_min;j<=x_max;j++) {
dsx=celldx[FTNREF1D(j,x_min-2)];
dsy=celldy[FTNREF1D(k,y_min-2)];
cc=soundspeed[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]*soundspeed[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)];
cc=cc+2.0*viscosity[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]/density0[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)];
cc=MAX(sqrt(cc),g_small);
dtct=dtc_safe*MIN(dsx,dsy)/cc;
div=0.0;
dv1=(xvel0[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]+xvel0[FTNREF2D(j ,k+1,x_max+5,x_min-2,y_min-2)])*xarea[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
dv2=(xvel0[FTNREF2D(j+1,k ,x_max+5,x_min-2,y_min-2)]+xvel0[FTNREF2D(j+1,k+1,x_max+5,x_min-2,y_min-2)])*xarea[FTNREF2D(j+1,k ,x_max+5,x_min-2,y_min-2)];
div=div+dv2-dv1;
dtut=dtu_safe*2.0*volume[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]/MAX(fabs(dv1),MAX(fabs(dv2),g_small*volume[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]));
dv1=(yvel0[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)]+yvel0[FTNREF2D(j+1,k,x_max+5,x_min-2,y_min-2)])*yarea[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)];
dv2=(yvel0[FTNREF2D(j,k+1,x_max+5,x_min-2,y_min-2)]+yvel0[FTNREF2D(j+1,k+1,x_max+5,x_min-2,y_min-2)])*yarea[FTNREF2D(j,k+1,x_max+4,x_min-2,y_min-2)];
div=div+dv2-dv1;
dtvt=dtv_safe*2.0*volume[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]/MAX(fabs(dv1),MAX(fabs(dv2),g_small*volume[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]));
div=div/(2.0*volume[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]);
if(div < -g_small) {
dtdivt=dtdiv_safe*(-1.0/div);
}
else {
dtdivt=g_big;
}
dt_min[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)]=MIN(dtct,MIN(dtut,MIN(dtvt,dtdivt)));
}
}
// Commenting out this one pragma - seems to be in Fortran format ('min')
//#pragma omp for private(j) reduction(min:dt_min_val)
for (k=y_min;k<=y_max;k++) {
#pragma ivdep
for (j=x_min;j<=x_max;j++) {
if(dt_min[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)] < dt_min_val) dt_min_val=dt_min[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)];
}
}
}
// Extract the mimimum timestep information
dtl_control=10.01*(jk_control-(int)(jk_control));
jk_control=jk_control-(jk_control-(int)(jk_control));
j_ldt=1; //MOD(INT(jk_control),x_max)
k_ldt=1; //1+(jk_control/x_max)
//xl_pos=cellx[FTNREF1D(j_ldt,xmin-2)];
//yl_pos=celly[FTNREF1D(j_ldt,ymin-2)];
if(dt_min_val < min_dt) small=1;
*dtminval=dt_min_val;
*dtlcontrol=1;
*xlpos=xl_pos;
*ylpos=yl_pos;
*jldt=j_ldt;
*kldt=k_ldt;
if(small != 0) {
printf("Timestep information:\n");
printf("j, k :%i %i \n",j_ldt,k_ldt);
printf("x, y :%f %f \n",xl_pos,yl_pos);
printf("timestep : %f\n",dt_min_val);
printf("Cell velocities;\n");
printf("%f %f \n",xvel0[FTNREF2D(j_ldt ,k_ldt ,x_max+5,x_min-2,y_min-2)],yvel0[FTNREF2D(j_ldt ,k_ldt ,x_max+5,x_min-2,y_min-2)]);
printf("%f %f \n",xvel0[FTNREF2D(j_ldt+1,k_ldt ,x_max+5,x_min-2,y_min-2)],yvel0[FTNREF2D(j_ldt+1,k_ldt ,x_max+5,x_min-2,y_min-2)]);
printf("%f %f \n",xvel0[FTNREF2D(j_ldt+1,k_ldt+1,x_max+5,x_min-2,y_min-2)],yvel0[FTNREF2D(j_ldt+1,k_ldt+1,x_max+5,x_min-2,y_min-2)]);
printf("%f %f \n",xvel0[FTNREF2D(j_ldt ,k_ldt+1,x_max+5,x_min-2,y_min-2)],yvel0[FTNREF2D(j_ldt ,k_ldt+1,x_max+5,x_min-2,y_min-2)]);
printf("density, energy, pressure, soundspeed \n");
printf("%f %f %f %f \n",density0[FTNREF2D(j_ldt,k_ldt,x_max+4,x_min-2,y_min-2)],energy0[FTNREF2D(j_ldt,k_ldt,x_max+4,x_min-2,y_min-2)],pressure[FTNREF2D(j_ldt,k_ldt,x_max+4,x_min-2,y_min-2)],soundspeed[FTNREF2D(j_ldt,k_ldt,x_max+4,x_min-2,y_min-2)]);
}
}
void initialise_chunk_kernel_c_(int *xmin,int *xmax,int *ymin,int *ymax,
double *minx,
double *miny,
double *dx,
double *dy,
double *vertexx,
double *vertexdx,
double *vertexy,
double *vertexdy,
double *cellx,
double *celldx,
double *celly,
double *celldy,
double *volume,
double *xarea,
double *yarea)
{
int x_min=*xmin;
int x_max=*xmax;
int y_min=*ymin;
int y_max=*ymax;
double min_x=*minx;
double min_y=*miny;
double d_x=*dx;
double d_y=*dy;
START_PROFILING;
#pragma omp parallel for
#pragma ivdep
for (int j = x_min-2; j <= x_max+3; j++)
{
vertexx[FTNREF1D(j,x_min-2)] =
min_x+d_x*(double)(j-x_min);
}
#pragma omp parallel for
#pragma ivdep
for (int j = x_min-2; j <= x_max+3; j++)
{
vertexdx[FTNREF1D(j,x_min-2)] = d_x;
}
#pragma omp parallel for
#pragma ivdep
for (int k = y_min-2; k <= y_max+3; k++)
{
vertexy[FTNREF1D(k,y_min-2)] =
min_y+d_y*(double)(k-y_min);
}
#pragma omp parallel for
#pragma ivdep
for (int k = y_min-2; k <= y_max+3; k++)
{
vertexdy[FTNREF1D(k,y_min-2)] = d_y;
}
#pragma omp parallel for
#pragma ivdep
for (int j = x_min-2; j <= x_max+2; j++)
{
cellx[FTNREF1D(j,x_min-2)] = 0.5 *
(vertexx[FTNREF1D(j,x_min-2)]+vertexx[FTNREF1D(j+1,x_min-2)]);
}
#pragma omp parallel for
#pragma ivdep
for (int j = x_min-2; j <= x_max+2; j++)
{
celldx[FTNREF1D(j,x_min-2)]=d_x;
}
#pragma omp parallel for
#pragma ivdep
for (int k = y_min-2; k <= y_max+2; k++)
{
celly[FTNREF1D(k,y_min-2)] = 0.5 *
(vertexy[FTNREF1D(k,y_min-2)]+vertexy[FTNREF1D(k+1,x_min-2)]);
}
#pragma omp parallel for
#pragma ivdep
for (int k = y_min-2; k <= y_max+2; k++)
{
celldy[FTNREF1D(k,y_min-2)] = d_y;
}
#pragma omp parallel for
for (int k = y_min-2; k <= y_max+2; k++)
{
#pragma ivdep
for (int j = x_min-2; j <= x_max+2; j++)
{
volume[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)] = d_x*d_y;
}
}
#pragma omp parallel for
for (int k = y_min-2; k <= y_max+2; k++)
{
#pragma ivdep
for (int j = x_min-2; j <= x_max+2; j++)
{
xarea[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)]=celldy[FTNREF1D(k,y_min-2)];
}
}
#pragma omp parallel for
for (int k = y_min-2; k <= y_max+2; k++)
{
#pragma ivdep
for (int j = x_min-2; j <= x_max+2; j++)
{
yarea[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)] =
celldx[FTNREF1D(j,x_min-2)];
}
}
STOP_PROFILING(__func__);
}
void advec_cell_kernel_c_(int *g_mic_device,int *xmin,int *xmax,int *ymin,int *ymax,
int *dr,
int *swp_nmbr,
int *vctr,
double *vertexdx,
double *vertexdy,
double *volume,
double *density1,
double *energy1,
double *mass_flux_x,
double *vol_flux_x,
double *mass_flux_y,
double *vol_flux_y,
double *pre_vol,
double *post_vol,
double *pre_mass,
double *post_mass,
double *advec_vol,
double *post_ener,
double *ener_flux)
{
int x_min=*xmin;
int x_max=*xmax;
int y_min=*ymin;
int y_max=*ymax;
int sweep_number=*swp_nmbr;
int dir=*dr;
int vector=*vctr;
int j,k,upwind,donor,downwind,dif;
int g_xdir=1,g_ydir=2;
double sigma,sigmat,sigmav,sigmam,sigma3,sigma4,diffuw,diffdw,limiter;
double one_by_six;
one_by_six=1.0/6.0;
#pragma offload target(mic:*g_mic_device) \
in(density1 :length(0) alloc_if(0) free_if(0)) \
in(energy1 :length(0) alloc_if(0) free_if(0)) \
in(vol_flux_x :length(0) alloc_if(0) free_if(0)) \
in(vol_flux_y :length(0) alloc_if(0) free_if(0)) \
in(volume :length(0) alloc_if(0) free_if(0)) \
in(mass_flux_x :length(0) alloc_if(0) free_if(0)) \
in(mass_flux_y :length(0) alloc_if(0) free_if(0)) \
in(vertexdx :length(0) alloc_if(0) free_if(0)) \
in(vertexdy :length(0) alloc_if(0) free_if(0)) \
in(pre_vol :length(0) alloc_if(0) free_if(0)) \
in(post_vol :length(0) alloc_if(0) free_if(0)) \
in(post_ener :length(0) alloc_if(0) free_if(0)) \
in(pre_mass :length(0) alloc_if(0) free_if(0)) \
in(post_mass :length(0) alloc_if(0) free_if(0)) \
in(advec_vol :length(0) alloc_if(0) free_if(0)) \
in(ener_flux :length(0) alloc_if(0) free_if(0))
#pragma omp parallel
{
if(dir==g_xdir){
if(sweep_number==1){
#pragma omp for private(j)
for (k=y_min-2;k<=y_max+2;k++) {
#pragma ivdep
for (j=x_min-2;j<=x_max+2;j++) {
pre_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=volume[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
+(vol_flux_x[FTNREF2D(j+1,k ,x_max+5,x_min-2,y_min-2)]
-vol_flux_x[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
+vol_flux_y[FTNREF2D(j ,k+1,x_max+4,x_min-2,y_min-2)]
-vol_flux_y[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]);
post_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=pre_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
-(vol_flux_x[FTNREF2D(j+1,k ,x_max+5,x_min-2,y_min-2)]
-vol_flux_x[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]);
}
}
}
else {
#pragma omp for private(j)
for (k=y_min-2;k<=y_max+2;k++) {
#pragma ivdep
for (j=x_min-2;j<=x_max+2;j++) {
pre_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=volume[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
+vol_flux_x[FTNREF2D(j+1,k ,x_max+5,x_min-2,y_min-2)]
-vol_flux_x[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
post_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=volume[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
}
}
}
#pragma omp for private(upwind,donor,downwind,dif,sigmat,sigma3,sigma4,sigmav,sigma,sigmam,diffuw,diffdw,limiter,j)
for (k=y_min;k<=y_max;k++) {
for (j=x_min;j<=x_max+2;j++) {
if(vol_flux_x[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]>0.0){
upwind =j-2;
donor =j-1;
downwind =j;
dif =donor;
}
else {
upwind =MIN(j+1,x_max+2);
donor =j;
downwind =j-1;
dif =upwind;
}
sigmat=fabs(vol_flux_x[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)]/pre_vol[FTNREF2D(donor,k ,x_max+5,x_min-2,y_min-2)]);
sigma3=(1.0+sigmat)*(vertexdx[FTNREF1D(j,x_min-2)]/vertexdx[FTNREF1D(dif,x_min-2)]);
sigma4=2.0-sigmat;
sigma=sigmat;
sigmav=sigmat;
diffuw=density1[FTNREF2D(donor,k ,x_max+4,x_min-2,y_min-2)]-density1[FTNREF2D(upwind,k ,x_max+4,x_min-2,y_min-2)];
diffdw=density1[FTNREF2D(downwind,k ,x_max+4,x_min-2,y_min-2)]-density1[FTNREF2D(donor,k ,x_max+4,x_min-2,y_min-2)];
if(diffuw*diffdw>0.0){
limiter=(1.0-sigmav)*SIGN(1.0,diffdw)*MIN(fabs(diffuw),MIN(fabs(diffdw)
,one_by_six*(sigma3*fabs(diffuw)+sigma4*fabs(diffdw))));
}
else{
limiter=0.0;
}
mass_flux_x[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)]=vol_flux_x[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)]
*(density1[FTNREF2D(donor,k ,x_max+4,x_min-2,y_min-2)]+limiter);
sigmam=fabs(mass_flux_x[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)])/(density1[FTNREF2D(donor,k ,x_max+4,x_min-2,y_min-2)]
*pre_vol[FTNREF2D(donor,k ,x_max+5,x_min-2,y_min-2)]);
diffuw=energy1[FTNREF2D(donor,k ,x_max+4,x_min-2,y_min-2)]-energy1[FTNREF2D(upwind,k ,x_max+4,x_min-2,y_min-2)];
diffdw=energy1[FTNREF2D(downwind,k ,x_max+4,x_min-2,y_min-2)]-energy1[FTNREF2D(donor,k ,x_max+4,x_min-2,y_min-2)];
if(diffuw*diffdw>0.0){
limiter=(1.0-sigmam)*SIGN(1.0,diffdw)*MIN(fabs(diffuw),MIN(fabs(diffdw)
,one_by_six*(sigma3*fabs(diffuw)+sigma4*fabs(diffdw))));
}
else {
limiter=0.0;
}
ener_flux[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)]=mass_flux_x[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)]
*(energy1[FTNREF2D(donor,k ,x_max+4,x_min-2,y_min-2)]+limiter);
}
}
#pragma omp for private(j)
for (k=y_min;k<=y_max;k++) {
#pragma ivdep
for (j=x_min;j<=x_max;j++) {
pre_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=density1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
*pre_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
post_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=pre_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
+mass_flux_x[FTNREF2D(j ,k,x_max+5,x_min-2,y_min-2)]
-mass_flux_x[FTNREF2D(j+1,k,x_max+5,x_min-2,y_min-2)];
post_ener[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=(energy1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
*pre_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
+ener_flux[FTNREF2D(j ,k,x_max+5,x_min-2,y_min-2)]
-ener_flux[FTNREF2D(j+1,k,x_max+5,x_min-2,y_min-2)])
/post_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
advec_vol [FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=pre_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
+vol_flux_x[FTNREF2D(j ,k,x_max+5,x_min-2,y_min-2)]
-vol_flux_x[FTNREF2D(j+1,k,x_max+5,x_min-2,y_min-2)];
density1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]=post_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]/advec_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
energy1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]=post_ener[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
}
}
}
else if(dir==g_ydir){
if(sweep_number==1){
#pragma omp for private(j)
for (k=y_min-2;k<=y_max+2;k++) {
#pragma ivdep
for (j=x_min-2;j<=x_max+2;j++) {
pre_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=volume[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
+(vol_flux_y[FTNREF2D(j ,k+1,x_max+4,x_min-2,y_min-2)]
-vol_flux_y[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
+vol_flux_x[FTNREF2D(j+1,k ,x_max+5,x_min-2,y_min-2)]
-vol_flux_x[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]);
post_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=pre_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
-(vol_flux_y[FTNREF2D(j ,k+1,x_max+4,x_min-2,y_min-2)]
-vol_flux_y[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]);
}
}
}
else {
#pragma omp for private(j)
for (k=y_min-2;k<=y_max+2;k++) {
#pragma ivdep
for (j=x_min-2;j<=x_max+2;j++) {
pre_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=volume[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
+vol_flux_y[FTNREF2D(j ,k+1,x_max+4,x_min-2,y_min-2)]
-vol_flux_y[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
post_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=volume[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)];
}
}
}
#pragma omp for private(upwind,donor,downwind,dif,sigmat,sigma3,sigma4,sigmav,sigma,sigmam,diffuw,diffdw,limiter,j)
for (k=y_min;k<=y_max+2;k++) {
for (j=x_min;j<=x_max;j++) {
if(vol_flux_y[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]>0.0){
upwind =k-2;
donor =k-1;
downwind =k;
dif =donor;
}
else {
upwind =MIN(k+1,y_max+2);
donor =k;
downwind =k-1;
dif =upwind;
}
sigmat=fabs(vol_flux_y[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]/pre_vol[FTNREF2D(j ,donor,x_max+5,x_min-2,y_min-2)]);
sigma3=(1.0+sigmat)*(vertexdy[FTNREF1D(k,y_min-2)]/vertexdy[FTNREF1D(dif,y_min-2)]);
sigma4=2.0-sigmat;
sigma=sigmat;
sigmav=sigmat;
diffuw=density1[FTNREF2D(j ,donor,x_max+4,x_min-2,y_min-2)]-density1[FTNREF2D(j ,upwind,x_max+4,x_min-2,y_min-2)];
diffdw=density1[FTNREF2D(j ,downwind,x_max+4,x_min-2,y_min-2)]-density1[FTNREF2D(j ,donor,x_max+4,x_min-2,y_min-2)];
if(diffuw*diffdw>0.0){
limiter=(1.0-sigmav)*SIGN(1.0,diffdw)*MIN(fabs(diffuw),MIN(fabs(diffdw)
,one_by_six*(sigma3*fabs(diffuw)+sigma4*fabs(diffdw))));
}
else{
limiter=0.0;
}
mass_flux_y[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]=vol_flux_y[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]
*(density1[FTNREF2D(j ,donor,x_max+4,x_min-2,y_min-2)]+limiter);
sigmam=fabs(mass_flux_y[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)])/(density1[FTNREF2D(j ,donor,x_max+4,x_min-2,y_min-2)]
*pre_vol[FTNREF2D(j ,donor,x_max+5,x_min-2,y_min-2)]);
diffuw=energy1[FTNREF2D(j ,donor,x_max+4,x_min-2,y_min-2)]-energy1[FTNREF2D(j ,upwind,x_max+4,x_min-2,y_min-2)];
diffdw=energy1[FTNREF2D(j ,downwind,x_max+4,x_min-2,y_min-2)]-energy1[FTNREF2D(j ,donor,x_max+4,x_min-2,y_min-2)];
if(diffuw*diffdw>0.0){
limiter=(1.0-sigmam)*SIGN(1.0,diffdw)*MIN(fabs(diffuw),MIN(fabs(diffdw)
,one_by_six*(sigma3*fabs(diffuw)+sigma4*fabs(diffdw))));
}
else {
limiter=0.0;
}
ener_flux[FTNREF2D(j,k,x_max+5,x_min-2,y_min-2)]=mass_flux_y[FTNREF2D(j,k,x_max+4,x_min-2,y_min-2)]
*(energy1[FTNREF2D(j ,donor,x_max+4,x_min-2,y_min-2)]+limiter);
}
}
#pragma omp for private(j)
for (k=y_min;k<=y_max;k++) {
#pragma ivdep
for (j=x_min;j<=x_max;j++) {
pre_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=density1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
*pre_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
post_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=pre_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
+mass_flux_y[FTNREF2D(j,k ,x_max+4,x_min-2,y_min-2)]
-mass_flux_y[FTNREF2D(j,k+1,x_max+4,x_min-2,y_min-2)];
post_ener[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=(energy1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]
*pre_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
+ener_flux[FTNREF2D(j,k ,x_max+5,x_min-2,y_min-2)]
-ener_flux[FTNREF2D(j,k+1,x_max+5,x_min-2,y_min-2)])
/post_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
advec_vol [FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]=pre_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]
+vol_flux_y[FTNREF2D(j,k ,x_max+4,x_min-2,y_min-2)]
-vol_flux_y[FTNREF2D(j,k+1,x_max+4,x_min-2,y_min-2)];
density1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]=post_mass[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)]/advec_vol[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
energy1[FTNREF2D(j ,k ,x_max+4,x_min-2,y_min-2)]=post_ener[FTNREF2D(j ,k ,x_max+5,x_min-2,y_min-2)];
}
}
}
}
}