| /petsc/src/ksp/pc/impls/bjacobi/bjkokkos/ |
| H A D | bjkokkoskernels.kokkos.cxx | 155 …Scalar *glb_Aaa, const PetscInt team_size, MatInfo info, const PetscInt batch_sz, PCFailedReason *… in PCApply_BJKOKKOSKERNELS() argument 170 if (Nsolves_team > batch_sz) Nsolves_team = batch_sz; // silently fix this in PCApply_BJKOKKOSKERNELS() 274 for (PetscInt dmIdx = 0, s = 0, head = 0; dmIdx < jac->num_dms; dmIdx += batch_sz) { in PCApply_BJKOKKOSKERNELS() 278 …for (int bid = 0; bid < batch_sz; bid++) PetscCall(PetscPrintf(PetscObjectComm((PetscObject)A), "%… in PCApply_BJKOKKOSKERNELS() 282 for (int bid = 0; bid < batch_sz; bid++) { in PCApply_BJKOKKOSKERNELS() 288 head += batch_sz * jac->dm_Nf[dmIdx]; in PCApply_BJKOKKOSKERNELS() 314 if (batch_sz != 1) in PCApply_BJKOKKOSKERNELS() 315 …? "converged" : "diverged", handle.get_iteration_host(blkID), blkID % batch_sz, blkID / batch_sz)); in PCApply_BJKOKKOSKERNELS() 321 …ch %d did not converge with %d iterations\n", (int)(blkID / batch_sz), (int)blkID % batch_sz, hand… in PCApply_BJKOKKOSKERNELS() 325 if (batch_sz != 1) in PCApply_BJKOKKOSKERNELS() [all …]
|
| H A D | bjkokkos.kokkos.cxx | 550 …PetscInt batch_sz; // the number of repeated DMs, [DM_e_1, DM_e_2, DM_e_b… in PCApply_BJKOKKOS() local 587 batch_sz = *pNf; // number of times to repeat the DMs in PCApply_BJKOKKOS() 588 } else batch_sz = 1; in PCApply_BJKOKKOS() 589 …ck(nBlk % batch_sz == 0, PetscObjectComm((PetscObject)pc), PETSC_ERR_ARG_WRONG, "batch_sz = %" Pet… in PCApply_BJKOKKOS() 593 …ERNELS(pc, glb_bdata, glb_xdata, glb_Aai, glb_Aaj, glb_Aaa, team_size, info, batch_sz, &pcreason)); in PCApply_BJKOKKOS() 672 …if (batch_sz != 1) PetscCall(PetscPrintf(PetscObjectComm((PetscObject)A), "%s: max iterations per … in PCApply_BJKOKKOS() 675 for (PetscInt dmIdx = 0, head = 0, s = 0; dmIdx < jac->num_dms; dmIdx += batch_sz) { in PCApply_BJKOKKOS() 677 for (int bid = 0; bid < batch_sz; bid++) { in PCApply_BJKOKKOS() 693 …for (int bid = 0; bid < batch_sz; bid++) PetscCall(PetscPrintf(PetscObjectComm((PetscObject)A), "%… in PCApply_BJKOKKOS() 699 head += batch_sz * jac->dm_Nf[dmIdx]; in PCApply_BJKOKKOS() [all …]
|
| /petsc/src/ts/tests/ |
| H A D | ex30.c | 284 for (PetscInt v_id_0 = 0; v_id_0 < ctx->batch_sz; v_id_0 += numthreads) { in gridToParticles_private() 285 …t_FMT ": map back to particles\n", global_vertex_id + 1, num_vertices, v_id_0 + 1, ctx->batch_sz)); in gridToParticles_private() 381 …ertex_id_0 || printCtx->v_target >= printCtx->global_vertex_id_0 + ctx->batch_sz) PetscFunctionRet… in PostStep() 384 v_id = printCtx->v_target % ctx->batch_sz; in PostStep() 447 …PetscCall(VecViewFromOptions(printCtx->globXArray[LAND_PACK_IDX(v_id % ctx->batch_sz, grid)], NULL… in PostStep() 450 …PetscCall(VecViewFromOptions(printCtx->globXArray[LAND_PACK_IDX(v_id % ctx->batch_sz, grid + 1)], … in PostStep() 482 …batch_sz % numthreads == 0, PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, "batch size (-dm_landau_ba… in go() 484 … PetscInt_FMT " global items (vertices) %d DMs\n", ctx->num_grids, ctx->batch_sz, num_vertices, (i… in go() 534 …id_0 = 0; global_vertex_id_0 < num_vertices; global_vertex_id_0 += ctx->batch_sz, shift /= 2) { //… in go() 540 if (v_target >= global_vertex_id_0 && v_target < global_vertex_id_0 + ctx->batch_sz) { in go() [all …]
|
| /petsc/src/ts/utils/dmplexlandau/kokkos/ |
| H A D | landau.kokkos.cxx | 137 …icDataSet(DM plex, const PetscInt Nq, const PetscInt Nb, const PetscInt batch_sz, const PetscInt n… in LandauKokkosStaticDataSet() argument 246 …auto ipfdf_data = new Kokkos::View<PetscReal ***, Kokkos::LayoutLeft>("fdf", batch_sz, dim + 1, IP… in LandauKokkosStaticDataSet() 248 …auto ipfdf_data = new Kokkos::View<PetscReal ***, Kokkos::LayoutRight>("fdf", batch_sz, dim + 1, I… in LandauKokkosStaticDataSet() 381 …acobian(DM plex[], const PetscInt Nq, const PetscInt Nb, const PetscInt batch_sz, const PetscInt n… in LandauKokkosJacobian() argument 395 …const PetscInt coo_sz_batch = SData_d->coo_size / batch_sz; … in LandauKokkosJacobian() 481 …Kokkos::View<PetscScalar ****, Kokkos::LayoutRight> d_elem_mats("element matrices", batch_sz, num_… in LandauKokkosJacobian() 499 closure_sz *= batch_sz; in LandauKokkosJacobian() 512 …_max=%d\n", scr_bytes_fdf, KOKKOS_SHARED_LEVEL, (int)(num_cells_batch * batch_sz), team_size, vect… in LandauKokkosJacobian() 514 …"f, df", Kokkos::TeamPolicy<>(num_cells_batch * batch_sz, team_size, vector_size).set_scratch_size… in LandauKokkosJacobian() 521 const PetscInt moffset = LAND_MOFFSET(b_id, grid, batch_sz, num_grids, d_mat_offset); in LandauKokkosJacobian() [all …]
|
| /petsc/src/ts/utils/dmplexlandau/ |
| H A D | plexland.c | 112 for (PetscInt i = 0; i < ctx->num_grids * ctx->batch_sz; i++) subJ[i] = NULL; in LandauFormJacobian_Internal() 115 for (PetscInt tid = 0; tid < ctx->batch_sz; tid++) { in LandauFormJacobian_Internal() 156 PetscCall(PetscMalloc1(cellClosure_sz * ctx->batch_sz, &cellClosure)); in LandauFormJacobian_Internal() 162 for (PetscInt b_id = 0; b_id < ctx->batch_sz; b_id++) { // OpenMP (once) in LandauFormJacobian_Internal() 180 …batch_sz, PETSC_COMM_SELF, PETSC_ERR_PLIB, "iteration wrong %" PetscCount_FMT " != cellClosure_sz … in LandauFormJacobian_Internal() 204 …PetscCall(LandauKokkosJacobian(ctx->plex, Nq, Nb, ctx->batch_sz, ctx->num_grids, numCells, Eq_m, c… in LandauFormJacobian_Internal() 232 IPf_sz_tot = IPf_sz_glb * ctx->batch_sz; in LandauFormJacobian_Internal() 243 …for (PetscInt tid = 0; tid < ctx->batch_sz * elem_offset[num_grids]; tid++) { … in LandauFormJacobian_Internal() 250 …const PetscInt moffset = LAND_MOFFSET(b_id, grid, ctx->batch_sz, ctx->num_grids, ctx->mat_offset);… in LandauFormJacobian_Internal() 313 … for (PetscInt tid = 0; tid < ctx->batch_sz * elem_offset[num_grids]; tid++) { // for each element in LandauFormJacobian_Internal() [all …]
|
| /petsc/include/ |
| H A D | petsclandau.h | 191 PetscInt batch_sz; member 200 #define LAND_PACK_IDX(_b, _g) (_g * ctx->batch_sz + _b)
|
| /petsc/src/ts/utils/dmplexlandau/tutorials/ |
| H A D | ex2.c | 163 …batch_sz, PETSC_COMM_SELF, PETSC_ERR_PLIB, "nDMs != ctx->num_grids*ctx->batch_sz %" PetscInt_FMT "… in testSpitzer() 395 … PetscCall(DMCompositeGetAccessArray(pack, F, ctx->num_grids * ctx->batch_sz, NULL, globFarray)); in FormSource() 401 …PetscCall(DMCompositeRestoreAccessArray(pack, F, ctx->num_grids * ctx->batch_sz, NULL, globFarray)… in FormSource() 422 PetscCall(DMCompositeGetAccessArray(pack, X, ctx->num_grids * ctx->batch_sz, NULL, globXArray)); in Monitor() 455 if (reason && ctx->verbose > 3 && ctx->batch_sz > 1) { in Monitor() 461 for (PetscInt i = 0; i < ctx->batch_sz; i++) { in Monitor() 477 … PetscCall(DMCompositeRestoreAccessArray(pack, X, ctx->num_grids * ctx->batch_sz, NULL, globXArray… in Monitor()
|
| H A D | ex1.c | 373 for (PetscInt b_id = 0; b_id < ctx->batch_sz; b_id++) { in main()
|