// user function
inline void drhouupdx_kernel(const double *rhou_new, const double *rho_new,
const double *rhoE_new, double *rhou_res) {
double fni =
rhou_new[OPS_ACC0(0)] * rhou_new[OPS_ACC0(0)] / rho_new[OPS_ACC1(0)];
double p = gam1 * (rhoE_new[OPS_ACC2(0)] - 0.5 * fni);
fni = fni + p;
double fnim1 =
rhou_new[OPS_ACC0(-1)] * rhou_new[OPS_ACC0(-1)] / rho_new[OPS_ACC1(-1)];
p = gam1 * (rhoE_new[OPS_ACC2(-1)] - 0.5 * fnim1);
fnim1 = fnim1 + p;
double fnim2 =
rhou_new[OPS_ACC0(-2)] * rhou_new[OPS_ACC0(-2)] / rho_new[OPS_ACC1(-2)];
p = gam1 * (rhoE_new[OPS_ACC2(-2)] - 0.5 * fnim2);
fnim2 = fnim2 + p;
double fnip1 =
rhou_new[OPS_ACC0(1)] * rhou_new[OPS_ACC0(1)] / rho_new[OPS_ACC1(1)];
p = gam1 * (rhoE_new[OPS_ACC2(1)] - 0.5 * fnip1);
fnip1 = fnip1 + p;
double fnip2 =
rhou_new[OPS_ACC0(2)] * rhou_new[OPS_ACC0(2)] / rho_new[OPS_ACC1(2)];
p = gam1 * (rhoE_new[OPS_ACC2(2)] - 0.5 * fnip2);
fnip2 = fnip2 + p;
double deriv = (fnim2 - fnip2 + 8.0 * (fnip1 - fnim1)) / (12.00 * dx);
rhou_res[OPS_ACC3(0)] = deriv;
}
inline void advec_cell_kernel2_zdir( double *pre_vol, double *post_vol, const double *volume,
const double *vol_flux_z) {
pre_vol[OPS_ACC0(0,0,0)] = volume[OPS_ACC2(0,0,0)] + vol_flux_z[OPS_ACC3(0,0,1)] - vol_flux_z[OPS_ACC3(0,0,0)];
post_vol[OPS_ACC1(0,0,0)] = volume[OPS_ACC2(0,0,0)];
}
// user function
inline void flux_calc_kernelx(double *vol_flux_x, const double *xarea,
const double *xvel0, const double *xvel1) {
vol_flux_x[OPS_ACC0(0, 0)] =
0.25 * dt * (xarea[OPS_ACC1(0, 0)]) *
((xvel0[OPS_ACC2(0, 0)]) + (xvel0[OPS_ACC2(0, 1)]) +
(xvel1[OPS_ACC3(0, 0)]) + (xvel1[OPS_ACC3(0, 1)]));
}
//user function
inline
void flux_calc_kernelz( double *vol_flux_z, const double *zarea,
const double *zvel0, const double *zvel1) {
vol_flux_z[OPS_ACC0(0,0,0)] = 0.125 * dt * (zarea[OPS_ACC1(0,0,0)]) *
( zvel0[OPS_ACC2(0,0,0)] + zvel0[OPS_ACC2(1,0,0)] + zvel0[OPS_ACC2(1,0,0)] + zvel0[OPS_ACC2(1,1,0)] +
zvel1[OPS_ACC3(0,0,0)] + zvel1[OPS_ACC3(1,0,0)] + zvel1[OPS_ACC3(0,1,0)] + zvel1[OPS_ACC3(1,1,0)]);
}
// user function
inline void tea_leaf_common_init_diag_init_kernel(double *Mi, const double *Kx,
const double *Ky,
const double *rx,
const double *ry) {
Mi[OPS_ACC0(0, 0)] =
1.0 / (1.0 + (*ry) * (Ky[OPS_ACC2(0, 1)] + Ky[OPS_ACC2(0, 0)]) +
(*rx) * (Kx[OPS_ACC1(1, 0)] + Kx[OPS_ACC1(0, 0)]));
}
// user function
inline void calc_dt_kernel(const double *celldx, const double *celldy,
const double *soundspeed, const double *viscosity,
const double *density0, const double *xvel0,
const double *xarea, const double *volume,
const double *yvel0, const double *yarea,
double *dt_min, const double *celldz,
const double *zvel0, const double *zarea) {
double div, ds, dtut, dtvt, dtct, dtwt, dtdivt, cc, dv1, dv2, du1, du2, dw1,
dw2;
ds = MIN(MIN(celldx[OPS_ACC0(0, 0, 0)], celldy[OPS_ACC1(0, 0, 0)]),
celldz[OPS_ACC11(0, 0, 0)]);
ds = 1.0 / (ds * ds);
cc = soundspeed[OPS_ACC2(0, 0, 0)] * soundspeed[OPS_ACC2(0, 0, 0)];
cc = cc + 2.0 * viscosity[OPS_ACC3(0, 0, 0)] / density0[OPS_ACC4(0, 0, 0)];
dtct = ds * cc;
dtct = dtc_safe * 1.0 / MAX(sqrt(dtct), g_small);
du1 = (xvel0[OPS_ACC5(0, 0, 0)] + xvel0[OPS_ACC5(0, 1, 0)] +
xvel0[OPS_ACC5(0, 0, 1)] + xvel0[OPS_ACC5(0, 1, 1)]) *
xarea[OPS_ACC6(0, 0, 0)];
du2 = (xvel0[OPS_ACC5(1, 0, 0)] + xvel0[OPS_ACC5(1, 1, 0)] +
xvel0[OPS_ACC5(1, 0, 1)] + xvel0[OPS_ACC5(1, 1, 1)]) *
xarea[OPS_ACC6(0, 0, 0)];
dtut = dtu_safe * 4.0 * volume[OPS_ACC7(0, 0, 0)] /
MAX(MAX(fabs(du1), fabs(du2)), 1.0e-5 * volume[OPS_ACC7(0, 0, 0)]);
dv1 = (yvel0[OPS_ACC8(0, 0, 0)] + yvel0[OPS_ACC8(1, 0, 0)] +
yvel0[OPS_ACC8(0, 0, 1)] + yvel0[OPS_ACC8(1, 0, 1)]) *
yarea[OPS_ACC9(0, 0, 0)];
dv2 = (yvel0[OPS_ACC8(0, 1, 0)] + yvel0[OPS_ACC8(1, 1, 0)] +
yvel0[OPS_ACC8(0, 1, 1)] + yvel0[OPS_ACC8(1, 1, 1)]) *
yarea[OPS_ACC9(0, 0, 0)];
dtvt = dtv_safe * 4.0 * volume[OPS_ACC7(0, 0, 0)] /
MAX(MAX(fabs(dv1), fabs(dv2)), 1.0e-5 * volume[OPS_ACC7(0, 0, 0)]);
dw1 = (zvel0[OPS_ACC12(0, 0, 0)] + zvel0[OPS_ACC12(0, 1, 0)] +
zvel0[OPS_ACC12(1, 0, 0)] + zvel0[OPS_ACC12(1, 1, 0)]) *
zarea[OPS_ACC13(0, 0, 0)];
dw2 = (zvel0[OPS_ACC12(0, 0, 1)] + zvel0[OPS_ACC12(0, 1, 1)] +
zvel0[OPS_ACC12(1, 0, 1)] + zvel0[OPS_ACC12(1, 1, 1)]) *
zarea[OPS_ACC13(0, 0, 0)];
dtwt = dtw_safe * 4.0 * volume[OPS_ACC7(0, 0, 0)] /
MAX(MAX(fabs(dw1), fabs(dw2)), 1.0e-5 * volume[OPS_ACC7(0, 0, 0)]);
div = du2 - du1 + dv2 - dv1 + dw2 - dw1;
dtdivt = dtdiv_safe * 4.0 * (volume[OPS_ACC7(0, 0, 0)]) /
MAX(volume[OPS_ACC7(0, 0, 0)] * 1.0e-05, fabs(div));
dt_min[OPS_ACC10(0, 0, 0)] =
MIN(MIN(MIN(dtct, dtut), MIN(dtvt, dtdivt)), dtwt);
}
// user function
inline void flux_calc_kernely(double *vol_flux_y, const double *yarea,
const double *yvel0, const double *yvel1) {
vol_flux_y[OPS_ACC0(0, 0, 0)] =
0.125 * dt * (yarea[OPS_ACC1(0, 0, 0)]) *
(yvel0[OPS_ACC2(0, 0, 0)] + yvel0[OPS_ACC2(1, 0, 0)] +
yvel0[OPS_ACC2(0, 0, 1)] + yvel0[OPS_ACC2(1, 0, 1)] +
yvel1[OPS_ACC3(0, 0, 0)] + yvel1[OPS_ACC3(1, 0, 0)] +
yvel1[OPS_ACC3(0, 0, 1)] + yvel1[OPS_ACC3(1, 0, 1)]);
}
// user function
inline void calvar_kernel(const double *rho_new, const double *rhou_new,
const double *rhoE_new, double *workarray2,
double *workarray3) {
double p, rhoi, u;
rhoi = 1 / rho_new[OPS_ACC0(0)];
u = rhou_new[OPS_ACC1(0)] * rhoi;
p = gam1 * (rhoE_new[OPS_ACC2(0)] - 0.5 * rho_new[OPS_ACC0(0)] * u * u);
workarray2[OPS_ACC3(0)] = p + rhou_new[OPS_ACC1(0)] * u;
workarray3[OPS_ACC4(0)] = (p + rhoE_new[OPS_ACC2(0)]) * u;
}
inline void advec_mom_kernel_post_pre_advec_y( double *node_mass_post, const double *post_vol,
const double *density1, double *node_mass_pre, const double *node_flux) {
node_mass_post[OPS_ACC0(0,0)] = 0.25 * ( density1[OPS_ACC2(0,-1)] * post_vol[OPS_ACC1(0,-1)] +
density1[OPS_ACC2(0,0)] * post_vol[OPS_ACC1(0,0)] +
density1[OPS_ACC2(-1,-1)] * post_vol[OPS_ACC1(-1,-1)] +
density1[OPS_ACC2(-1,0)] * post_vol[OPS_ACC1(-1,0)] );
node_mass_pre[OPS_ACC3(0,0)] = node_mass_post[OPS_ACC0(0,0)] - node_flux[OPS_ACC4(0,-1)] + node_flux[OPS_ACC4(0,0)];
}
inline void update_halo_kernel1_fr2(double *density0, double *density1,
double *energy0, double *energy1,
double *pressure, double *viscosity,
double *soundspeed , const int* fields) {
if(fields[FIELD_DENSITY0] == 1) density0[OPS_ACC0(0,0,0)] = density0[OPS_ACC0(0,0,-3)];
if(fields[FIELD_DENSITY1] == 1) density1[OPS_ACC1(0,0,0)] = density1[OPS_ACC1(0,0,-3)];
if(fields[FIELD_ENERGY0] == 1) energy0[OPS_ACC2(0,0,0)] = energy0[OPS_ACC2(0,0,-3)];
if(fields[FIELD_ENERGY1] == 1) energy1[OPS_ACC3(0,0,0)] = energy1[OPS_ACC3(0,0,-3)];
if(fields[FIELD_PRESSURE] == 1) pressure[OPS_ACC4(0,0,0)] = pressure[OPS_ACC4(0,0,-3)];
if(fields[FIELD_VISCOSITY] == 1) viscosity[OPS_ACC5(0,0,0)] = viscosity[OPS_ACC5(0,0,-3)];
if(fields[FIELD_SOUNDSPEED] == 1) soundspeed[OPS_ACC6(0,0,0)] = soundspeed[OPS_ACC6(0,0,-3)];
}
// user function
inline void tea_leaf_cg_calc_w_reduce_kernel(double *w, const double *Kx,
const double *Ky, const double *p,
const double *rx, const double *ry,
double *pw) {
w[OPS_ACC0(0, 0)] = (1.0 + (*ry) * (Ky[OPS_ACC2(0, 1)] + Ky[OPS_ACC2(0, 0)]) +
(*rx) * (Kx[OPS_ACC1(1, 0)] + Kx[OPS_ACC1(0, 0)])) *
p[OPS_ACC3(0, 0)] -
(*ry) * (Ky[OPS_ACC2(0, 1)] * p[OPS_ACC3(0, 1)] +
Ky[OPS_ACC2(0, 0)] * p[OPS_ACC3(0, -1)]) -
(*rx) * (Kx[OPS_ACC1(1, 0)] * p[OPS_ACC3(1, 0)] +
Kx[OPS_ACC1(0, 0)] * p[OPS_ACC3(-1, 0)]);
*pw = *pw + w[OPS_ACC0(0, 0)] * p[OPS_ACC3(0, 0)];
}
// user function
inline void tea_leaf_common_init_Kx_Ky_kernel(double *Kx, double *Ky,
const double *w) {
Kx[OPS_ACC0(0, 0)] = (w[OPS_ACC2(-1, 0)] + w[OPS_ACC2(0, 0)]) /
(2.0 * w[OPS_ACC2(-1, 0)] * w[OPS_ACC2(0, 0)]);
Ky[OPS_ACC1(0, 0)] = (w[OPS_ACC2(0, -1)] + w[OPS_ACC2(0, 0)]) /
(2.0 * w[OPS_ACC2(0, -1)] * w[OPS_ACC2(0, 0)]);
}
// user function
inline void calc_dt_kernel_print(const double *xvel0, const double *yvel0,
const double *zvel0, const double *density0,
const double *energy0, const double *pressure,
const double *soundspeed, double *output) {
output[0] = xvel0[OPS_ACC0(0, 0, 0)];
output[1] = yvel0[OPS_ACC1(0, 0, 0)];
output[2] = zvel0[OPS_ACC2(0, 0, 0)];
output[3] = xvel0[OPS_ACC0(1, 0, 0)];
output[4] = yvel0[OPS_ACC1(1, 0, 0)];
output[5] = zvel0[OPS_ACC2(0, 0, 0)];
output[6] = xvel0[OPS_ACC0(1, 1, 0)];
output[7] = yvel0[OPS_ACC1(1, 1, 0)];
output[8] = zvel0[OPS_ACC2(0, 0, 0)];
output[9] = xvel0[OPS_ACC0(0, 1, 0)];
output[10] = yvel0[OPS_ACC1(0, 1, 0)];
output[11] = zvel0[OPS_ACC2(0, 0, 0)];
output[12] = xvel0[OPS_ACC0(0, 0, 1)];
output[13] = yvel0[OPS_ACC1(0, 0, 1)];
output[14] = zvel0[OPS_ACC2(0, 0, 1)];
output[15] = xvel0[OPS_ACC0(1, 0, 1)];
output[16] = yvel0[OPS_ACC1(1, 0, 1)];
output[17] = zvel0[OPS_ACC2(0, 0, 1)];
output[18] = xvel0[OPS_ACC0(1, 1, 1)];
output[19] = yvel0[OPS_ACC1(1, 1, 1)];
output[20] = zvel0[OPS_ACC2(0, 0, 1)];
output[21] = xvel0[OPS_ACC0(0, 1, 1)];
output[22] = yvel0[OPS_ACC1(0, 1, 1)];
output[23] = zvel0[OPS_ACC2(0, 0, 1)];
output[24] = density0[OPS_ACC3(0, 0, 0)];
output[25] = energy0[OPS_ACC4(0, 0, 0)];
output[26] = pressure[OPS_ACC5(0, 0, 0)];
output[27] = soundspeed[OPS_ACC6(0, 0, 0)];
}
inline void update_halo_kernel1_r2(double *density0, double *energy0,
double *energy1, double *u, double *p,
double *sd, const int *fields) {
if (fields[FIELD_DENSITY] == 1)
density0[OPS_ACC0(0, 0)] = density0[OPS_ACC0(-3, 0)];
if (fields[FIELD_ENERGY0] == 1)
energy0[OPS_ACC1(0, 0)] = energy0[OPS_ACC1(-3, 0)];
if (fields[FIELD_ENERGY1] == 1)
energy1[OPS_ACC2(0, 0)] = energy1[OPS_ACC2(-3, 0)];
if (fields[FIELD_U] == 1)
u[OPS_ACC3(0, 0)] = u[OPS_ACC3(-3, 0)];
if (fields[FIELD_P] == 1)
p[OPS_ACC4(0, 0)] = p[OPS_ACC4(-3, 0)];
if (fields[FIELD_SD] == 1)
sd[OPS_ACC5(0, 0)] = sd[OPS_ACC5(-3, 0)];
}
//user function
inline void preproc_kernel(const double *u, double *du,
double *ax, double *bx, double *cx, double *ay, double *by, double *cy,
double *az, double *bz, double *cz, int *idx){
double a, b, c, d;
if(idx[0]==0 || idx[0]==nx-1 || idx[1]==0 || idx[1]==ny-1 || idx[2]==0 || idx[2]==nz-1) {
d = 0.0f;
a = 0.0f;
b = 1.0f;
c = 0.0f;
} else {
d = lambda*( u[OPS_ACC0(-1,0,0)] + u[OPS_ACC0(1,0,0)]
+ u[OPS_ACC0(0,-1,0)] + u[OPS_ACC0(0,1,0)]
+ u[OPS_ACC0(0,0,-1)] + u[OPS_ACC0(0,0,1)]
- 6.0f*u[OPS_ACC0(0,0,0)]);
a = -0.5f * lambda;
b = 1.0f + lambda;
c = -0.5f * lambda;
}
du[OPS_ACC1(0,0,0)] = d;
ax[OPS_ACC2(0,0,0)] = a;
bx[OPS_ACC3(0,0,0)] = b;
cx[OPS_ACC4(0,0,0)] = c;
ay[OPS_ACC5(0,0,0)] = a;
by[OPS_ACC6(0,0,0)] = b;
cy[OPS_ACC7(0,0,0)] = c;
az[OPS_ACC8(0,0,0)] = a;
bz[OPS_ACC9(0,0,0)] = b;
cz[OPS_ACC10(0,0,0)] = c;
}
// user function
inline void save_kernel(double *rho_old, double *rhou_old, double *rhoE_old,
const double *rho_new, const double *rhou_new,
const double *rhoE_new) {
rho_old[OPS_ACC0(0)] = rho_new[OPS_ACC3(0)];
rhou_old[OPS_ACC1(0)] = rhou_new[OPS_ACC4(0)];
rhoE_old[OPS_ACC2(0)] = rhoE_new[OPS_ACC5(0)];
}
// user function
inline void field_summary_kernel(const double *volume, const double *density0,
const double *energy0, const double *pressure,
const double *xvel0, const double *yvel0,
double *vol, double *mass, double *ie,
double *ke, double *press) {
double vsqrd, cell_vol, cell_mass;
vsqrd = 0.0;
vsqrd = vsqrd +
0.25 * (xvel0[OPS_ACC4(0, 0)] * xvel0[OPS_ACC4(0, 0)] +
yvel0[OPS_ACC5(0, 0)] * yvel0[OPS_ACC5(0, 0)]);
vsqrd = vsqrd +
0.25 * (xvel0[OPS_ACC4(1, 0)] * xvel0[OPS_ACC4(1, 0)] +
yvel0[OPS_ACC5(1, 0)] * yvel0[OPS_ACC5(1, 0)]);
vsqrd = vsqrd +
0.25 * (xvel0[OPS_ACC4(0, 1)] * xvel0[OPS_ACC4(0, 1)] +
yvel0[OPS_ACC5(0, 1)] * yvel0[OPS_ACC5(0, 1)]);
vsqrd = vsqrd +
0.25 * (xvel0[OPS_ACC4(1, 1)] * xvel0[OPS_ACC4(1, 1)] +
yvel0[OPS_ACC5(1, 1)] * yvel0[OPS_ACC5(1, 1)]);
cell_vol = volume[OPS_ACC0(0, 0)];
cell_mass = cell_vol * density0[OPS_ACC1(0, 0)];
*vol = *vol + cell_vol;
*mass = *mass + cell_mass;
*ie = *ie + cell_mass * energy0[OPS_ACC2(0, 0)];
*ke = *ke + cell_mass * 0.5 * vsqrd;
*press = *press + cell_vol * pressure[OPS_ACC3(0, 0)];
}
// user function
inline void ideal_gas_kernel(const double *density, const double *energy,
double *pressure, double *soundspeed) {
double sound_speed_squared, v, pressurebyenergy, pressurebyvolume;
v = 1.0 / density[OPS_ACC0(0, 0, 0)];
pressure[OPS_ACC2(0, 0, 0)] =
(1.4 - 1.0) * density[OPS_ACC0(0, 0, 0)] * energy[OPS_ACC1(0, 0, 0)];
pressurebyenergy = (1.4 - 1.0) * density[OPS_ACC0(0, 0, 0)];
pressurebyvolume =
-1.0 * density[OPS_ACC0(0, 0, 0)] * pressure[OPS_ACC2(0, 0, 0)];
sound_speed_squared =
v * v *
(pressure[OPS_ACC2(0, 0, 0)] * pressurebyenergy - pressurebyvolume);
soundspeed[OPS_ACC3(0, 0, 0)] = sqrt(sound_speed_squared);
}
// user function
inline void calc_dt_kernel(const double *celldx, const double *celldy,
const double *soundspeed, const double *viscosity,
const double *density0, const double *xvel0,
const double *xarea, const double *volume,
const double *yvel0, const double *yarea,
double *dt_min) {
double div, dsx, dsy, dtut, dtvt, dtct, dtdivt, cc, dv1, dv2;
dsx = celldx[OPS_ACC0(0, 0)];
dsy = celldy[OPS_ACC1(0, 0)];
cc = soundspeed[OPS_ACC2(0, 0)] * soundspeed[OPS_ACC2(0, 0)];
cc = cc + 2.0 * viscosity[OPS_ACC3(0, 0)] / density0[OPS_ACC4(0, 0)];
cc = MAX(sqrt(cc), g_small);
dtct = dtc_safe * MIN(dsx, dsy) / cc;
div = 0.0;
dv1 = (xvel0[OPS_ACC5(0, 0)] + xvel0[OPS_ACC5(0, 1)]) * xarea[OPS_ACC6(0, 0)];
dv2 = (xvel0[OPS_ACC5(1, 0)] + xvel0[OPS_ACC5(1, 1)]) * xarea[OPS_ACC6(1, 0)];
div = div + dv2 - dv1;
dtut = dtu_safe * 2.0 * volume[OPS_ACC7(0, 0)] /
MAX(MAX(fabs(dv1), fabs(dv2)), g_small * volume[OPS_ACC7(0, 0)]);
dv1 = (yvel0[OPS_ACC8(0, 0)] + yvel0[OPS_ACC8(1, 0)]) * yarea[OPS_ACC9(0, 0)];
dv2 = (yvel0[OPS_ACC8(0, 1)] + yvel0[OPS_ACC8(1, 1)]) * yarea[OPS_ACC9(0, 1)];
div = div + dv2 - dv1;
dtvt = dtv_safe * 2.0 * volume[OPS_ACC7(0, 0)] /
MAX(MAX(fabs(dv1), fabs(dv2)), g_small * volume[OPS_ACC7(0, 0)]);
div = div / (2.0 * volume[OPS_ACC7(0, 0)]);
if (div < -g_small)
dtdivt = dtdiv_safe * (-1.0 / div);
else
dtdivt = g_big;
dt_min[OPS_ACC10(0, 0)] = MIN(MIN(dtct, dtut), MIN(dtvt, dtdivt));
}
inline void advec_mom_kernel_x1( double *pre_vol, double *post_vol,
const double *volume,
const double *vol_flux_x, const double *vol_flux_y, const double *vol_flux_z) {
post_vol[OPS_ACC1(0,0,0)] = volume[OPS_ACC2(0,0,0)] + vol_flux_y[OPS_ACC4(0,1,0)] - vol_flux_y[OPS_ACC4(0,0,0)]
+ vol_flux_z[OPS_ACC5(0,0,1)] - vol_flux_z[OPS_ACC5(0,0,0)];
pre_vol[OPS_ACC0(0,0,0)] = post_vol[OPS_ACC1(0,0,0)] + vol_flux_x[OPS_ACC3(1,0,0)] - vol_flux_x[OPS_ACC3(0,0,0)];
}
// user function
inline void advec_mom_kernel2_x(double *vel1, const double *node_mass_post,
const double *node_mass_pre,
const double *mom_flux) {
vel1[OPS_ACC0(0, 0, 0)] =
(vel1[OPS_ACC0(0, 0, 0)] * node_mass_pre[OPS_ACC2(0, 0, 0)] +
mom_flux[OPS_ACC3(-1, 0, 0)] - mom_flux[OPS_ACC3(0, 0, 0)]) /
node_mass_post[OPS_ACC1(0, 0, 0)];
}
inline void advec_mom_kernel_y2(double *pre_vol, double *post_vol,
const double *volume,
const double *vol_flux_x) {
post_vol[OPS_ACC1(0, 0)] = volume[OPS_ACC2(0, 0)];
pre_vol[OPS_ACC0(0, 0)] = post_vol[OPS_ACC1(0, 0)] +
vol_flux_x[OPS_ACC3(1, 0)] -
vol_flux_x[OPS_ACC3(0, 0)];
}
// user function
inline void initialise_chunk_kernel_celly(const double *vertexy, double *celly,
double *celldy) {
double d_y;
d_y = (grid.ymax - grid.ymin) / (double)grid.y_cells;
celly[OPS_ACC1(0, 0)] =
0.5 * (vertexy[OPS_ACC0(0, 0)] + vertexy[OPS_ACC0(0, 1)]);
celldy[OPS_ACC2(0, 0)] = d_y;
}
// user function
inline void initialise_chunk_kernel_y(double *vertexy, const int *yy,
double *vertexdy) {
int y_min = field.y_min - 2;
double min_y, d_y;
d_y = (grid.ymax - grid.ymin) / (double)grid.y_cells;
min_y = grid.ymin + d_y * field.bottom;
vertexy[OPS_ACC0(0, 0, 0)] = min_y + d_y * (yy[OPS_ACC1(0, 0, 0)] - y_min);
vertexdy[OPS_ACC2(0, 0, 0)] = (double)d_y;
}
// user function
inline void checkop_kernel(const double *rho_new, const double *x,
const double *rhoin, double *pre, double *post,
int *num) {
double diff;
diff = (rho_new[OPS_ACC0(0)] - rhoin[OPS_ACC2(0)]);
if (fabs(diff) < 0.01 && x[OPS_ACC1(0)] > -4.1) {
*post = *post + diff * diff;
*num = *num + 1;
} else
*pre = *pre + (rho_new[OPS_ACC0(0)] - rhol) * (rho_new[OPS_ACC0(0)] - rhol);
}
// user function
inline void advec_cell_kernel1_ydir(double *pre_vol, double *post_vol,
const double *volume,
const double *vol_flux_x,
const double *vol_flux_y) {
pre_vol[OPS_ACC0(0, 0)] =
volume[OPS_ACC2(0, 0)] +
(vol_flux_y[OPS_ACC4(0, 1)] - vol_flux_y[OPS_ACC4(0, 0)] +
vol_flux_x[OPS_ACC3(1, 0)] - vol_flux_x[OPS_ACC3(0, 0)]);
post_vol[OPS_ACC1(0, 0)] =
pre_vol[OPS_ACC0(0, 0)] -
(vol_flux_y[OPS_ACC4(0, 1)] - vol_flux_y[OPS_ACC4(0, 0)]);
}
// user function
inline void initialise_chunk_kernel_volume(double *volume, const double *celldy,
double *xarea, const double *celldx,
double *yarea) {
double d_x, d_y;
d_x = (grid.xmax - grid.xmin) / (double)grid.x_cells;
d_y = (grid.ymax - grid.ymin) / (double)grid.y_cells;
volume[OPS_ACC0(0, 0)] = d_x * d_y;
xarea[OPS_ACC2(0, 0)] = celldy[OPS_ACC1(0, 0)];
yarea[OPS_ACC4(0, 0)] = celldx[OPS_ACC3(0, 0)];
}
// user function
inline void
advec_cell_kernel4_zdir(double *density1, double *energy1,
const double *mass_flux_z, const double *vol_flux_z,
const double *pre_vol, const double *post_vol,
double *pre_mass, double *post_mass, double *advec_vol,
double *post_ener, const double *ener_flux) {
pre_mass[OPS_ACC6(0, 0, 0)] =
density1[OPS_ACC0(0, 0, 0)] * pre_vol[OPS_ACC4(0, 0, 0)];
post_mass[OPS_ACC7(0, 0, 0)] = pre_mass[OPS_ACC6(0, 0, 0)] +
mass_flux_z[OPS_ACC2(0, 0, 0)] -
mass_flux_z[OPS_ACC2(0, 0, 1)];
post_ener[OPS_ACC9(0, 0, 0)] =
(energy1[OPS_ACC1(0, 0, 0)] * pre_mass[OPS_ACC6(0, 0, 0)] +
ener_flux[OPS_ACC10(0, 0, 0)] - ener_flux[OPS_ACC10(0, 0, 1)]) /
post_mass[OPS_ACC7(0, 0, 0)];
advec_vol[OPS_ACC8(0, 0, 0)] = pre_vol[OPS_ACC4(0, 0, 0)] +
vol_flux_z[OPS_ACC3(0, 0, 0)] -
vol_flux_z[OPS_ACC3(0, 0, 1)];
density1[OPS_ACC0(0, 0, 0)] =
post_mass[OPS_ACC7(0, 0, 0)] / advec_vol[OPS_ACC8(0, 0, 0)];
energy1[OPS_ACC1(0, 0, 0)] = post_ener[OPS_ACC9(0, 0, 0)];
}
// user function
inline void Riemann_kernel(const double *rho_new, const double *rhou_new,
const double *rhoE_new, double *alam, double *r,
double *al) {
double rl, rr, rho, u, hl, hr, h, Vsq, csq, c;
double dw1, dw2, dw3, delpc2, rdeluc;
rl = sqrt(rho_new[OPS_ACC0(0)]);
rr = sqrt(rho_new[OPS_ACC0(1)]);
rho = rl + rr;
u = ((rhou_new[OPS_ACC1(0)] / rl) + (rhou_new[OPS_ACC1(1)] / rr)) / rho;
double fni =
rhou_new[OPS_ACC1(0)] * rhou_new[OPS_ACC1(0)] / rho_new[OPS_ACC0(0)];
double p = gam1 * (rhoE_new[OPS_ACC2(0)] - 0.5 * fni);
hl = (rhoE_new[OPS_ACC2(0)] + p) / rl;
fni = rhou_new[OPS_ACC1(1)] * rhou_new[OPS_ACC1(1)] / rho_new[OPS_ACC0(1)];
p = gam1 * (rhoE_new[OPS_ACC2(1)] - 0.5 * fni);
hr = (rhoE_new[OPS_ACC2(1)] + p) / rr;
h = (hl + hr) / rho;
Vsq = u * u;
csq = gam1 * (h - 0.5 * Vsq);
c = sqrt(csq);
alam[OPS_ACC_MD3(0, 0)] = u - c;
alam[OPS_ACC_MD3(1, 0)] = u;
alam[OPS_ACC_MD3(2, 0)] = u + c;
r[OPS_ACC_MD4(0, 0)] = 1.0;
r[OPS_ACC_MD4(1, 0)] = 1.0;
r[OPS_ACC_MD4(2, 0)] = 1.0;
r[OPS_ACC_MD4(3, 0)] = u - c;
r[OPS_ACC_MD4(4, 0)] = u;
r[OPS_ACC_MD4(5, 0)] = u + c;
r[OPS_ACC_MD4(6, 0)] = h - u * c;
r[OPS_ACC_MD4(7, 0)] = 0.5 * Vsq;
r[OPS_ACC_MD4(8, 0)] = h + u * c;
for (int m = 0; m < 9; m++)
r[OPS_ACC_MD4(m, 0)] = r[OPS_ACC_MD4(m, 0)] / csq;
dw1 = rho_new[OPS_ACC0(1)] - rho_new[OPS_ACC0(0)];
dw2 = rhou_new[OPS_ACC1(1)] - rhou_new[OPS_ACC1(0)];
dw3 = rhoE_new[OPS_ACC2(1)] - rhoE_new[OPS_ACC2(0)];
delpc2 = gam1 * (dw3 + 0.50 * Vsq * dw1 - u * dw2) / csq;
rdeluc = (dw2 - u * dw1) / c;
al[OPS_ACC_MD5(0, 0)] = 0.5 * (delpc2 - rdeluc);
al[OPS_ACC_MD5(1, 0)] = dw1 - delpc2;
al[OPS_ACC_MD5(2, 0)] = 0.5 * (delpc2 + rdeluc);
for (int m = 0; m < 3; m++)
al[OPS_ACC_MD5(m, 0)] = al[OPS_ACC_MD5(m, 0)] * csq;
}
// user function
inline void calc_dt_kernel_print(const double *xvel0, const double *yvel0,
const double *density0, const double *energy0,
const double *pressure,
const double *soundspeed, double *output) {
output[0] = xvel0[OPS_ACC0(1, 0)];
output[1] = yvel0[OPS_ACC1(1, 0)];
output[2] = xvel0[OPS_ACC0(-1, 0)];
output[3] = yvel0[OPS_ACC1(-1, 0)];
output[4] = xvel0[OPS_ACC0(0, 1)];
output[5] = yvel0[OPS_ACC1(0, 1)];
output[6] = xvel0[OPS_ACC0(0, -1)];
output[7] = yvel0[OPS_ACC1(0, -1)];
output[8] = density0[OPS_ACC2(0, 0)];
output[9] = energy0[OPS_ACC3(0, 0)];
output[10] = pressure[OPS_ACC4(0, 0)];
output[11] = soundspeed[OPS_ACC5(0, 0)];
}
