diff options
Diffstat (limited to 'ML_ADM/src/ML_ADM_RHS.cc')
| -rw-r--r-- | ML_ADM/src/ML_ADM_RHS.cc | 148 |
1 files changed, 42 insertions, 106 deletions
diff --git a/ML_ADM/src/ML_ADM_RHS.cc b/ML_ADM/src/ML_ADM_RHS.cc index 4aa5793..b24ebe3 100644 --- a/ML_ADM/src/ML_ADM_RHS.cc +++ b/ML_ADM/src/ML_ADM_RHS.cc @@ -50,8 +50,6 @@ static void ML_ADM_RHS_Body(cGH const * restrict const cctkGH, int const dir, in DECLARE_CCTK_PARAMETERS; - /* Declare finite differencing variables */ - /* Include user-supplied include files */ /* Initialise finite differencing variables */ @@ -82,30 +80,30 @@ static void ML_ADM_RHS_Body(cGH const * restrict const cctkGH, int const dir, in CCTK_REAL_VEC const p1o12dx = kmul(INV(dx),ToReal(0.0833333333333333333333333333333)); CCTK_REAL_VEC const p1o12dy = kmul(INV(dy),ToReal(0.0833333333333333333333333333333)); CCTK_REAL_VEC const p1o12dz = kmul(INV(dz),ToReal(0.0833333333333333333333333333333)); - CCTK_REAL_VEC const p1o144dxdy = kmul(INV(dx),kmul(INV(dy),ToReal(0.00694444444444444444444444444444))); - CCTK_REAL_VEC const p1o144dxdz = kmul(INV(dx),kmul(INV(dz),ToReal(0.00694444444444444444444444444444))); - CCTK_REAL_VEC const p1o144dydz = kmul(INV(dy),kmul(INV(dz),ToReal(0.00694444444444444444444444444444))); + CCTK_REAL_VEC const p1o144dxdy = kmul(INV(kmul(dx,dy)),ToReal(0.00694444444444444444444444444444)); + CCTK_REAL_VEC const p1o144dxdz = kmul(INV(kmul(dx,dz)),ToReal(0.00694444444444444444444444444444)); + CCTK_REAL_VEC const p1o144dydz = kmul(INV(kmul(dy,dz)),ToReal(0.00694444444444444444444444444444)); CCTK_REAL_VEC const p1o180dx2 = kmul(INV(SQR(dx)),ToReal(0.00555555555555555555555555555556)); CCTK_REAL_VEC const p1o180dy2 = kmul(INV(SQR(dy)),ToReal(0.00555555555555555555555555555556)); CCTK_REAL_VEC const p1o180dz2 = kmul(INV(SQR(dz)),ToReal(0.00555555555555555555555555555556)); CCTK_REAL_VEC const p1o2dx = kmul(INV(dx),ToReal(0.5)); CCTK_REAL_VEC const p1o2dy = kmul(INV(dy),ToReal(0.5)); CCTK_REAL_VEC const p1o2dz = kmul(INV(dz),ToReal(0.5)); - CCTK_REAL_VEC const p1o3600dxdy = kmul(INV(dx),kmul(INV(dy),ToReal(0.000277777777777777777777777777778))); - CCTK_REAL_VEC const p1o3600dxdz = kmul(INV(dx),kmul(INV(dz),ToReal(0.000277777777777777777777777777778))); - CCTK_REAL_VEC const p1o3600dydz = kmul(INV(dy),kmul(INV(dz),ToReal(0.000277777777777777777777777777778))); - CCTK_REAL_VEC const p1o4dxdy = kmul(INV(dx),kmul(INV(dy),ToReal(0.25))); - CCTK_REAL_VEC const p1o4dxdz = kmul(INV(dx),kmul(INV(dz),ToReal(0.25))); - CCTK_REAL_VEC const p1o4dydz = kmul(INV(dy),kmul(INV(dz),ToReal(0.25))); + CCTK_REAL_VEC const p1o3600dxdy = kmul(INV(kmul(dx,dy)),ToReal(0.000277777777777777777777777777778)); + CCTK_REAL_VEC const p1o3600dxdz = kmul(INV(kmul(dx,dz)),ToReal(0.000277777777777777777777777777778)); + CCTK_REAL_VEC const p1o3600dydz = kmul(INV(kmul(dy,dz)),ToReal(0.000277777777777777777777777777778)); + CCTK_REAL_VEC const p1o4dxdy = kmul(INV(kmul(dx,dy)),ToReal(0.25)); + CCTK_REAL_VEC const p1o4dxdz = kmul(INV(kmul(dx,dz)),ToReal(0.25)); + CCTK_REAL_VEC const p1o4dydz = kmul(INV(kmul(dy,dz)),ToReal(0.25)); CCTK_REAL_VEC const p1o5040dx2 = kmul(INV(SQR(dx)),ToReal(0.000198412698412698412698412698413)); CCTK_REAL_VEC const p1o5040dy2 = kmul(INV(SQR(dy)),ToReal(0.000198412698412698412698412698413)); CCTK_REAL_VEC const p1o5040dz2 = kmul(INV(SQR(dz)),ToReal(0.000198412698412698412698412698413)); CCTK_REAL_VEC const p1o60dx = kmul(INV(dx),ToReal(0.0166666666666666666666666666667)); CCTK_REAL_VEC const p1o60dy = kmul(INV(dy),ToReal(0.0166666666666666666666666666667)); CCTK_REAL_VEC const p1o60dz = kmul(INV(dz),ToReal(0.0166666666666666666666666666667)); - CCTK_REAL_VEC const p1o705600dxdy = kmul(INV(dx),kmul(INV(dy),ToReal(1.41723356009070294784580498866e-6))); - CCTK_REAL_VEC const p1o705600dxdz = kmul(INV(dx),kmul(INV(dz),ToReal(1.41723356009070294784580498866e-6))); - CCTK_REAL_VEC const p1o705600dydz = kmul(INV(dy),kmul(INV(dz),ToReal(1.41723356009070294784580498866e-6))); + CCTK_REAL_VEC const p1o705600dxdy = kmul(INV(kmul(dx,dy)),ToReal(1.41723356009070294784580498866e-6)); + CCTK_REAL_VEC const p1o705600dxdz = kmul(INV(kmul(dx,dz)),ToReal(1.41723356009070294784580498866e-6)); + CCTK_REAL_VEC const p1o705600dydz = kmul(INV(kmul(dy,dz)),ToReal(1.41723356009070294784580498866e-6)); CCTK_REAL_VEC const p1o840dx = kmul(INV(dx),ToReal(0.00119047619047619047619047619048)); CCTK_REAL_VEC const p1o840dy = kmul(INV(dy),ToReal(0.00119047619047619047619047619048)); CCTK_REAL_VEC const p1o840dz = kmul(INV(dz),ToReal(0.00119047619047619047619047619048)); @@ -171,7 +169,7 @@ static void ML_ADM_RHS_Body(cGH const * restrict const cctkGH, int const dir, in /* Loop over the grid points */ #pragma omp parallel - LC_LOOP3VEC (ML_ADM_RHS, + LC_LOOP3VEC(ML_ADM_RHS, i,j,k, imin[0],imin[1],imin[2], imax[0],imax[1],imax[2], cctk_lsh[0],cctk_lsh[1],cctk_lsh[2], CCTK_REAL_VEC_SIZE) @@ -1473,96 +1471,26 @@ static void ML_ADM_RHS_Body(cGH const * restrict const cctkGH, int const dir, in CCTK_REAL_VEC beta3rhsL = ToReal(0); - /* If necessary, store only partial vectors after the first iteration */ - - if (CCTK_REAL_VEC_SIZE > 2 && CCTK_BUILTIN_EXPECT(i < lc_imin && i+CCTK_REAL_VEC_SIZE > lc_imax, 0)) - { - ptrdiff_t const elt_count_lo = lc_imin-i; - ptrdiff_t const elt_count_hi = lc_imax-i; - vec_store_nta_partial_mid(alpharhs[index],alpharhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(beta1rhs[index],beta1rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(beta2rhs[index],beta2rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(beta3rhs[index],beta3rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(g11rhs[index],g11rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(g12rhs[index],g12rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(g13rhs[index],g13rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(g22rhs[index],g22rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(g23rhs[index],g23rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(g33rhs[index],g33rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(K11rhs[index],K11rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(K12rhs[index],K12rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(K13rhs[index],K13rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(K22rhs[index],K22rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(K23rhs[index],K23rhsL,elt_count_lo,elt_count_hi); - vec_store_nta_partial_mid(K33rhs[index],K33rhsL,elt_count_lo,elt_count_hi); - break; - } - - /* If necessary, store only partial vectors after the first iteration */ - - if (CCTK_REAL_VEC_SIZE > 1 && CCTK_BUILTIN_EXPECT(i < lc_imin, 0)) - { - ptrdiff_t const elt_count = lc_imin-i; - vec_store_nta_partial_hi(alpharhs[index],alpharhsL,elt_count); - vec_store_nta_partial_hi(beta1rhs[index],beta1rhsL,elt_count); - vec_store_nta_partial_hi(beta2rhs[index],beta2rhsL,elt_count); - vec_store_nta_partial_hi(beta3rhs[index],beta3rhsL,elt_count); - vec_store_nta_partial_hi(g11rhs[index],g11rhsL,elt_count); - vec_store_nta_partial_hi(g12rhs[index],g12rhsL,elt_count); - vec_store_nta_partial_hi(g13rhs[index],g13rhsL,elt_count); - vec_store_nta_partial_hi(g22rhs[index],g22rhsL,elt_count); - vec_store_nta_partial_hi(g23rhs[index],g23rhsL,elt_count); - vec_store_nta_partial_hi(g33rhs[index],g33rhsL,elt_count); - vec_store_nta_partial_hi(K11rhs[index],K11rhsL,elt_count); - vec_store_nta_partial_hi(K12rhs[index],K12rhsL,elt_count); - vec_store_nta_partial_hi(K13rhs[index],K13rhsL,elt_count); - vec_store_nta_partial_hi(K22rhs[index],K22rhsL,elt_count); - vec_store_nta_partial_hi(K23rhs[index],K23rhsL,elt_count); - vec_store_nta_partial_hi(K33rhs[index],K33rhsL,elt_count); - continue; - } - - /* If necessary, store only partial vectors after the last iteration */ - - if (CCTK_REAL_VEC_SIZE > 1 && CCTK_BUILTIN_EXPECT(i+CCTK_REAL_VEC_SIZE > lc_imax, 0)) - { - ptrdiff_t const elt_count = lc_imax-i; - vec_store_nta_partial_lo(alpharhs[index],alpharhsL,elt_count); - vec_store_nta_partial_lo(beta1rhs[index],beta1rhsL,elt_count); - vec_store_nta_partial_lo(beta2rhs[index],beta2rhsL,elt_count); - vec_store_nta_partial_lo(beta3rhs[index],beta3rhsL,elt_count); - vec_store_nta_partial_lo(g11rhs[index],g11rhsL,elt_count); - vec_store_nta_partial_lo(g12rhs[index],g12rhsL,elt_count); - vec_store_nta_partial_lo(g13rhs[index],g13rhsL,elt_count); - vec_store_nta_partial_lo(g22rhs[index],g22rhsL,elt_count); - vec_store_nta_partial_lo(g23rhs[index],g23rhsL,elt_count); - vec_store_nta_partial_lo(g33rhs[index],g33rhsL,elt_count); - vec_store_nta_partial_lo(K11rhs[index],K11rhsL,elt_count); - vec_store_nta_partial_lo(K12rhs[index],K12rhsL,elt_count); - vec_store_nta_partial_lo(K13rhs[index],K13rhsL,elt_count); - vec_store_nta_partial_lo(K22rhs[index],K22rhsL,elt_count); - vec_store_nta_partial_lo(K23rhs[index],K23rhsL,elt_count); - vec_store_nta_partial_lo(K33rhs[index],K33rhsL,elt_count); - break; - } - vec_store_nta(alpharhs[index],alpharhsL); - vec_store_nta(beta1rhs[index],beta1rhsL); - vec_store_nta(beta2rhs[index],beta2rhsL); - vec_store_nta(beta3rhs[index],beta3rhsL); - vec_store_nta(g11rhs[index],g11rhsL); - vec_store_nta(g12rhs[index],g12rhsL); - vec_store_nta(g13rhs[index],g13rhsL); - vec_store_nta(g22rhs[index],g22rhsL); - vec_store_nta(g23rhs[index],g23rhsL); - vec_store_nta(g33rhs[index],g33rhsL); - vec_store_nta(K11rhs[index],K11rhsL); - vec_store_nta(K12rhs[index],K12rhsL); - vec_store_nta(K13rhs[index],K13rhsL); - vec_store_nta(K22rhs[index],K22rhsL); - vec_store_nta(K23rhs[index],K23rhsL); - vec_store_nta(K33rhs[index],K33rhsL); + /* Copy local copies back to grid functions */ + vec_store_partial_prepare(i,lc_imin,lc_imax); + vec_store_nta_partial(alpharhs[index],alpharhsL); + vec_store_nta_partial(beta1rhs[index],beta1rhsL); + vec_store_nta_partial(beta2rhs[index],beta2rhsL); + vec_store_nta_partial(beta3rhs[index],beta3rhsL); + vec_store_nta_partial(g11rhs[index],g11rhsL); + vec_store_nta_partial(g12rhs[index],g12rhsL); + vec_store_nta_partial(g13rhs[index],g13rhsL); + vec_store_nta_partial(g22rhs[index],g22rhsL); + vec_store_nta_partial(g23rhs[index],g23rhsL); + vec_store_nta_partial(g33rhs[index],g33rhsL); + vec_store_nta_partial(K11rhs[index],K11rhsL); + vec_store_nta_partial(K12rhs[index],K12rhsL); + vec_store_nta_partial(K13rhs[index],K13rhsL); + vec_store_nta_partial(K22rhs[index],K22rhsL); + vec_store_nta_partial(K23rhs[index],K23rhsL); + vec_store_nta_partial(K33rhs[index],K33rhsL); } - LC_ENDLOOP3VEC (ML_ADM_RHS); + LC_ENDLOOP3VEC(ML_ADM_RHS); } extern "C" void ML_ADM_RHS(CCTK_ARGUMENTS) @@ -1581,7 +1509,15 @@ extern "C" void ML_ADM_RHS(CCTK_ARGUMENTS) return; } - const char *groups[] = {"ML_ADM::ML_curv","ML_ADM::ML_curvrhs","ML_ADM::ML_lapse","ML_ADM::ML_lapserhs","ML_ADM::ML_metric","ML_ADM::ML_metricrhs","ML_ADM::ML_shift","ML_ADM::ML_shiftrhs"}; + const char *const groups[] = { + "ML_ADM::ML_curv", + "ML_ADM::ML_curvrhs", + "ML_ADM::ML_lapse", + "ML_ADM::ML_lapserhs", + "ML_ADM::ML_metric", + "ML_ADM::ML_metricrhs", + "ML_ADM::ML_shift", + "ML_ADM::ML_shiftrhs"}; GenericFD_AssertGroupStorage(cctkGH, "ML_ADM_RHS", 8, groups); switch(fdOrder) @@ -1603,7 +1539,7 @@ extern "C" void ML_ADM_RHS(CCTK_ARGUMENTS) break; } - GenericFD_LoopOverInterior(cctkGH, &ML_ADM_RHS_Body); + GenericFD_LoopOverInterior(cctkGH, ML_ADM_RHS_Body); if (verbose > 1) { |