/* A simple 2D hydro code (C) Romain Teyssier : CEA/IRFU -- original F90 code (C) Pierre-Francois Lavallee : IDRIS -- original F90 code (C) Guillaume Colin de Verdiere : CEA/DAM -- for the C version */ /* This software is governed by the CeCILL license under French law and abiding by the rules of distribution of free software. You can use, modify and/ or redistribute the software under the terms of the CeCILL license as circulated by CEA, CNRS and INRIA at the following URL "http://www.cecill.info". As a counterpart to the access to the source code and rights to copy, modify and redistribute granted by the license, users are provided only with a limited warranty and the software's author, the holder of the economic rights, and the successive licensors have only limited liability. In this respect, the user's attention is drawn to the risks associated with loading, using, modifying and/or developing or reproducing the software by the user in light of its specific status of free software, that may mean that it is complicated to manipulate, and that also therefore means that it is reserved for developers and experienced professionals having in-depth computer knowledge. Users are therefore encouraged to load and test the software's suitability as regards their requirements in conditions enabling the security of their systems and/or data to be ensured and, more generally, to use and operate it in the same conditions as regards security. The fact that you are presently reading this means that you have had knowledge of the CeCILL license and that you accept its terms. */ #include #include #include #include #include "parametres.h" #include "perfcnt.h" #include "utils.h" #include "riemann.h" #ifdef HMPP #undef HMPP #include "constoprim.c" #include "equation_of_state.c" #include "slope.c" #include "trace.c" #include "qleftright.c" #include "cmpflx.c" #include "conservar.c" #define HMPP #endif #define PRECISION 1e-6 void Dmemset(size_t nbr, real_t t[nbr], real_t motif) { int i; for (i = 0; i < nbr; i++) { t[i] = motif; } } #define DABS(x) (real_t) fabs((x)) #ifdef HMPP #define MAX(x,y) fmax(x,y) #endif #define MYSQRT sqrt void riemann(int narray, const real_t Hsmallr, const real_t Hsmallc, const real_t Hgamma, const int Hniter_riemann, const int Hnvar, const int Hnxyt, const int slices, const int Hstep, real_t qleft[Hnvar][Hstep][Hnxyt], real_t qright[Hnvar][Hstep][Hnxyt], // real_t qgdnv[Hnvar][Hstep][Hnxyt], // int sgnm[Hstep][Hnxyt], hydrowork_t * Hw) { int i, s, ii, iimx; real_t smallp_ = Square(Hsmallc) / Hgamma; real_t gamma6_ = (Hgamma + one) / (two * Hgamma); real_t smallpp_ = Hsmallr * smallp_; FLOPS(4, 2, 0, 0); // __declspec(align(256)) thevariable int *Fgoon = Hw->goon; real_t *Fpstar = Hw->pstar; real_t *Frl = Hw->rl; real_t *Ful = Hw->ul; real_t *Fpl = Hw->pl; real_t *Fur = Hw->ur; real_t *Fpr = Hw->pr; real_t *Fcl = Hw->cl; real_t *Fcr = Hw->cr; real_t *Frr = Hw->rr; real_t smallp = smallp_; real_t gamma6 = gamma6_; real_t smallpp = smallpp_; // fprintf(stderr, "%d\n", __ICC ); #pragma message "active pragma simd " #define SIMDNEEDED 1 #if __ICC < 1300 #define SIMD ivdep #else #define SIMD simd #endif // #define SIMD novector // Pressure, density and velocity #pragma omp parallel for private(s, i), shared(qgdnv, sgnm) reduction(+:flopsAri), reduction(+:flopsSqr), reduction(+:flopsMin), reduction(+:flopsTra) for (s = 0; s < slices; s++) { int ii, iimx; int *goon; real_t *pstar, *rl, *ul, *pl, *rr, *ur, *pr, *cl, *cr; int iter; pstar = &Fpstar[s * narray]; rl = &Frl[s * narray]; ul = &Ful[s * narray]; pl = &Fpl[s * narray]; rr = &Frr[s * narray]; ur = &Fur[s * narray]; pr = &Fpr[s * narray]; cl = &Fcl[s * narray]; cr = &Fcr[s * narray]; goon = &Fgoon[s * narray]; // Precompute values for this slice #ifdef SIMDNEEDED #if __ICC < 1300 #pragma ivdep #else #pragma SIMD #endif #endif for (i = 0; i < narray; i++) { rl[i] = fmax(qleft[ID][s][i], Hsmallr); ul[i] = qleft[IU][s][i]; pl[i] = fmax(qleft[IP][s][i], (real_t) (rl[i] * smallp)); rr[i] = fmax(qright[ID][s][i], Hsmallr); ur[i] = qright[IU][s][i]; pr[i] = fmax(qright[IP][s][i], (real_t) (rr[i] * smallp)); // Lagrangian sound speed cl[i] = Hgamma * pl[i] * rl[i]; cr[i] = Hgamma * pr[i] * rr[i]; // First guess real_t wl_i = MYSQRT(cl[i]); real_t wr_i = MYSQRT(cr[i]); pstar[i] = fmax(((wr_i * pl[i] + wl_i * pr[i]) + wl_i * wr_i * (ul[i] - ur[i])) / (wl_i + wr_i), 0.0); goon[i] = 1; } #define Fmax(a,b) (((a) > (b)) ? (a): (b)) #define Fabs(a) (((a) > 0) ? (a): -(a)) // solve the riemann problem on the interfaces of this slice for (iter = 0; iter < Hniter_riemann; iter++) { #ifdef SIMDNEEDED #if __ICC < 1300 #pragma simd #else #pragma SIMD #endif #endif for (i = 0; i < narray; i++) { if (goon[i]) { real_t pst = pstar[i]; // Newton-Raphson iterations to find pstar at the required accuracy real_t wwl = MYSQRT(cl[i] * (one + gamma6 * (pst - pl[i]) / pl[i])); real_t wwr = MYSQRT(cr[i] * (one + gamma6 * (pst - pr[i]) / pr[i])); real_t swwl = Square(wwl); real_t ql = two * wwl * swwl / (swwl + cl[i]); real_t qr = two * wwr * Square(wwr) / (Square(wwr) + cr[i]); real_t usl = ul[i] - (pst - pl[i]) / wwl; real_t usr = ur[i] + (pst - pr[i]) / wwr; real_t tmp = (qr * ql / (qr + ql) * (usl - usr)); real_t delp_i = Fmax(tmp, (-pst)); // pstar[i] = pstar[i] + delp_i; pst += delp_i; // Convergence indicator real_t tmp2 = delp_i / (pst + smallpp); real_t uo_i = Fabs(tmp2); goon[i] = uo_i > PRECISION; // FLOPS(29, 10, 2, 0); pstar[i] = pst; } } } // iter_riemann #ifdef SIMDNEEDED #pragma SIMD #endif for (i = 0; i < narray; i++) { real_t wl_i = MYSQRT(cl[i]); real_t wr_i = MYSQRT(cr[i]); wr_i = MYSQRT(cr[i] * (one + gamma6 * (pstar[i] - pr[i]) / pr[i])); wl_i = MYSQRT(cl[i] * (one + gamma6 * (pstar[i] - pl[i]) / pl[i])); real_t ustar_i = half * (ul[i] + (pl[i] - pstar[i]) / wl_i + ur[i] - (pr[i] - pstar[i]) / wr_i); int left = ustar_i > 0; real_t ro_i, uo_i, po_i, wo_i; if (left) { sgnm[s][i] = 1; ro_i = rl[i]; uo_i = ul[i]; po_i = pl[i]; wo_i = wl_i; } else { sgnm[s][i] = -1; ro_i = rr[i]; uo_i = ur[i]; po_i = pr[i]; wo_i = wr_i; } real_t co_i = MYSQRT(fabs(Hgamma * po_i / ro_i)); co_i = fmax(Hsmallc, co_i); real_t rstar_i = ro_i / (one + ro_i * (po_i - pstar[i]) / Square(wo_i)); rstar_i = fmax(rstar_i, Hsmallr); real_t cstar_i = MYSQRT(fabs(Hgamma * pstar[i] / rstar_i)); cstar_i = fmax(Hsmallc, cstar_i); real_t spout_i = co_i - sgnm[s][i] * uo_i; real_t spin_i = cstar_i - sgnm[s][i] * ustar_i; real_t ushock_i = wo_i / ro_i - sgnm[s][i] * uo_i; if (pstar[i] >= po_i) { spin_i = ushock_i; spout_i = ushock_i; } real_t scr_i = fmax((real_t) (spout_i - spin_i), (real_t) (Hsmallc + fabs(spout_i + spin_i))); real_t frac_i = (one + (spout_i + spin_i) / scr_i) * half; frac_i = fmax(zero, (real_t) (fmin(one, frac_i))); int addSpout = spout_i < zero; int addSpin = spin_i > zero; // real_t originalQgdnv = !addSpout & !addSpin; real_t qgdnv_ID, qgdnv_IU, qgdnv_IP; if (addSpout) { qgdnv_ID = ro_i; qgdnv_IU = uo_i; qgdnv_IP = po_i; } else if (addSpin) { qgdnv_ID = rstar_i; qgdnv_IU = ustar_i; qgdnv_IP = pstar[i]; } else { qgdnv_ID = (frac_i * rstar_i + (one - frac_i) * ro_i); qgdnv_IU = (frac_i * ustar_i + (one - frac_i) * uo_i); qgdnv_IP = (frac_i * pstar[i] + (one - frac_i) * po_i); } qgdnv[ID][s][i] = qgdnv_ID; qgdnv[IU][s][i] = qgdnv_IU; qgdnv[IP][s][i] = qgdnv_IP; // transverse velocity if (left) { qgdnv[IV][s][i] = qleft[IV][s][i]; } else { qgdnv[IV][s][i] = qright[IV][s][i]; } } } { int nops = slices * narray; FLOPS(57 * nops, 17 * nops, 14 * nops, 0 * nops); } // other passive variables if (Hnvar > IP) { int invar; for (invar = IP + 1; invar < Hnvar; invar++) { for (s = 0; s < slices; s++) { #ifdef SIMDNEEDED #pragma SIMD #endif for (i = 0; i < narray; i++) { int left = (sgnm[s][i] == 1); qgdnv[invar][s][i] = qleft[invar][s][i] * left + qright[invar][s][i] * !left; } } } } } // riemann_vec //EOF