1 // Copyright (c) 2017, Lawrence Livermore National Security, LLC. Produced at 2 // the Lawrence Livermore National Laboratory. LLNL-CODE-734707. All Rights 3 // reserved. See files LICENSE and NOTICE for details. 4 // 5 // This file is part of CEED, a collection of benchmarks, miniapps, software 6 // libraries and APIs for efficient high-order finite element and spectral 7 // element discretizations for exascale applications. For more information and 8 // source code availability see http://github.com/ceed. 9 // 10 // The CEED research is supported by the Exascale Computing Project 17-SC-20-SC, 11 // a collaborative effort of two U.S. Department of Energy organizations (Office 12 // of Science and the National Nuclear Security Administration) responsible for 13 // the planning and preparation of a capable exascale ecosystem, including 14 // software, applications, hardware, advanced system engineering and early 15 // testbed platforms, in support of the nation's exascale computing imperative. 16 17 // libCEED + PETSc Example: CEED BPs 3-6 with Multigrid 18 // 19 // This example demonstrates a simple usage of libCEED with PETSc to solve the 20 // CEED BP benchmark problems, see http://ceed.exascaleproject.org/bps. 21 // 22 // The code uses higher level communication protocols in DMPlex. 23 // 24 // Build with: 25 // 26 // make multigrid [PETSC_DIR=</path/to/petsc>] [CEED_DIR=</path/to/libceed>] 27 // 28 // Sample runs: 29 // 30 // multigrid -problem bp3 31 // multigrid -problem bp4 -ceed /cpu/self 32 // multigrid -problem bp5 -ceed /cpu/occa 33 // multigrid -problem bp6 -ceed /gpu/cuda 34 // 35 //TESTARGS -ceed {ceed_resource} -test -problem bp3 -degree 3 36 37 /// @file 38 /// CEED BPs 1-6 multigrid example using PETSc 39 const char help[] = "Solve CEED BPs using p-multigrid with PETSc and DMPlex\n"; 40 41 #define multigrid 42 #include "setup.h" 43 44 int main(int argc, char **argv) { 45 PetscInt ierr; 46 MPI_Comm comm; 47 char filename[PETSC_MAX_PATH_LEN], 48 ceedresource[PETSC_MAX_PATH_LEN] = "/cpu/self"; 49 double my_rt_start, my_rt, rt_min, rt_max; 50 PetscInt degree = 3, qextra, *lsize, *xlsize, *gsize, dim = 3, fineLevel, 51 melem[3] = {3, 3, 3}, ncompu = 1, numlevels = degree, *leveldegrees; 52 PetscScalar *r; 53 PetscBool test_mode, benchmark_mode, read_mesh, write_solution; 54 DM *dm, dmorig; 55 SNES snesdummy; 56 KSP ksp; 57 PC pc; 58 Mat *matO, *matPR, matcoarse; 59 Vec *X, *Xloc, *mult, rhs, rhsloc; 60 UserO *userO; 61 UserProlongRestr *userPR; 62 Ceed ceed; 63 CeedData *ceeddata; 64 CeedVector rhsceed, target; 65 CeedQFunction qferror, qfrestrict, qfprolong; 66 CeedOperator operror; 67 bpType bpchoice; 68 coarsenType coarsen; 69 70 ierr = PetscInitialize(&argc, &argv, NULL, help); 71 if (ierr) return ierr; 72 comm = PETSC_COMM_WORLD; 73 74 // Parse command line options 75 ierr = PetscOptionsBegin(comm, NULL, "CEED BPs in PETSc", NULL); CHKERRQ(ierr); 76 bpchoice = CEED_BP3; 77 ierr = PetscOptionsEnum("-problem", 78 "CEED benchmark problem to solve", NULL, 79 bpTypes, (PetscEnum)bpchoice, (PetscEnum *)&bpchoice, 80 NULL); CHKERRQ(ierr); 81 ncompu = bpOptions[bpchoice].ncompu; 82 test_mode = PETSC_FALSE; 83 ierr = PetscOptionsBool("-test", 84 "Testing mode (do not print unless error is large)", 85 NULL, test_mode, &test_mode, NULL); CHKERRQ(ierr); 86 benchmark_mode = PETSC_FALSE; 87 ierr = PetscOptionsBool("-benchmark", 88 "Benchmarking mode (prints benchmark statistics)", 89 NULL, benchmark_mode, &benchmark_mode, NULL); 90 CHKERRQ(ierr); 91 write_solution = PETSC_FALSE; 92 ierr = PetscOptionsBool("-write_solution", 93 "Write solution for visualization", 94 NULL, write_solution, &write_solution, NULL); 95 CHKERRQ(ierr); 96 degree = test_mode ? 3 : 2; 97 ierr = PetscOptionsInt("-degree", "Polynomial degree of tensor product basis", 98 NULL, degree, °ree, NULL); CHKERRQ(ierr); 99 if (degree < 1) SETERRQ1(PETSC_COMM_WORLD, PETSC_ERR_ARG_OUTOFRANGE, 100 "-degree %D must be at least 1", degree); 101 qextra = bpOptions[bpchoice].qextra; 102 ierr = PetscOptionsInt("-qextra", "Number of extra quadrature points", 103 NULL, qextra, &qextra, NULL); CHKERRQ(ierr); 104 ierr = PetscOptionsString("-ceed", "CEED resource specifier", 105 NULL, ceedresource, ceedresource, 106 sizeof(ceedresource), NULL); CHKERRQ(ierr); 107 coarsen = COARSEN_UNIFORM; 108 ierr = PetscOptionsEnum("-coarsen", 109 "Coarsening strategy to use", NULL, 110 coarsenTypes, (PetscEnum)coarsen, 111 (PetscEnum *)&coarsen, NULL); CHKERRQ(ierr); 112 read_mesh = PETSC_FALSE; 113 ierr = PetscOptionsString("-mesh", "Read mesh from file", NULL, 114 filename, filename, sizeof(filename), &read_mesh); 115 CHKERRQ(ierr); 116 if (!read_mesh) { 117 PetscInt tmp = dim; 118 ierr = PetscOptionsIntArray("-cells","Number of cells per dimension", NULL, 119 melem, &tmp, NULL); CHKERRQ(ierr); 120 } 121 ierr = PetscOptionsEnd(); CHKERRQ(ierr); 122 123 // Setup DM 124 if (read_mesh) { 125 ierr = DMPlexCreateFromFile(PETSC_COMM_WORLD, filename, PETSC_TRUE, &dmorig); 126 CHKERRQ(ierr); 127 } else { 128 ierr = DMPlexCreateBoxMesh(PETSC_COMM_WORLD, dim, PETSC_FALSE, melem, NULL, 129 NULL, NULL, PETSC_TRUE,&dmorig); CHKERRQ(ierr); 130 } 131 132 { 133 DM dmDist = NULL; 134 PetscPartitioner part; 135 136 ierr = DMPlexGetPartitioner(dmorig, &part); CHKERRQ(ierr); 137 ierr = PetscPartitionerSetFromOptions(part); CHKERRQ(ierr); 138 ierr = DMPlexDistribute(dmorig, 0, NULL, &dmDist); CHKERRQ(ierr); 139 if (dmDist) { 140 ierr = DMDestroy(&dmorig); CHKERRQ(ierr); 141 dmorig = dmDist; 142 } 143 } 144 145 // Allocate arrays for PETSc objects for each level 146 switch (coarsen) { 147 case COARSEN_UNIFORM: 148 numlevels = degree; 149 break; 150 case COARSEN_LOGARITHMIC: 151 numlevels = ceil(log(degree)/log(2)) + 1; 152 break; 153 } 154 ierr = PetscMalloc1(numlevels, &leveldegrees); CHKERRQ(ierr); 155 fineLevel = numlevels - 1; 156 157 switch (coarsen) { 158 case COARSEN_UNIFORM: 159 for (int i=0; i<numlevels; i++) leveldegrees[i] = i + 1; 160 break; 161 case COARSEN_LOGARITHMIC: 162 for (int i=0; i<numlevels - 1; i++) leveldegrees[i] = pow(2,i); 163 leveldegrees[fineLevel] = degree; 164 break; 165 } 166 ierr = PetscMalloc1(numlevels, &dm); CHKERRQ(ierr); 167 ierr = PetscMalloc1(numlevels, &X); CHKERRQ(ierr); 168 ierr = PetscMalloc1(numlevels, &Xloc); CHKERRQ(ierr); 169 ierr = PetscMalloc1(numlevels, &mult); CHKERRQ(ierr); 170 ierr = PetscMalloc1(numlevels, &userO); CHKERRQ(ierr); 171 ierr = PetscMalloc1(numlevels, &userPR); CHKERRQ(ierr); 172 ierr = PetscMalloc1(numlevels, &matO); CHKERRQ(ierr); 173 ierr = PetscMalloc1(numlevels, &matPR); CHKERRQ(ierr); 174 ierr = PetscMalloc1(numlevels, &lsize); CHKERRQ(ierr); 175 ierr = PetscMalloc1(numlevels, &xlsize); CHKERRQ(ierr); 176 ierr = PetscMalloc1(numlevels, &gsize); CHKERRQ(ierr); 177 178 // Setup DM and Operator Mat Shells for each level 179 for (CeedInt i=0; i<numlevels; i++) { 180 // Create DM 181 ierr = DMClone(dmorig, &dm[i]); CHKERRQ(ierr); 182 ierr = SetupDMByDegree(dm[i], leveldegrees[i], ncompu, bpchoice); 183 CHKERRQ(ierr); 184 185 // Create vectors 186 ierr = DMCreateGlobalVector(dm[i], &X[i]); CHKERRQ(ierr); 187 ierr = VecGetLocalSize(X[i], &lsize[i]); CHKERRQ(ierr); 188 ierr = VecGetSize(X[i], &gsize[i]); CHKERRQ(ierr); 189 ierr = DMCreateLocalVector(dm[i], &Xloc[i]); CHKERRQ(ierr); 190 ierr = VecGetSize(Xloc[i], &xlsize[i]); CHKERRQ(ierr); 191 192 // Operator 193 ierr = PetscMalloc1(1, &userO[i]); CHKERRQ(ierr); 194 ierr = MatCreateShell(comm, lsize[i], lsize[i], gsize[i], gsize[i], 195 userO[i], &matO[i]); CHKERRQ(ierr); 196 ierr = MatShellSetOperation(matO[i], MATOP_MULT, 197 (void(*)(void))MatMult_Ceed); CHKERRQ(ierr); 198 ierr = MatShellSetOperation(matO[i], MATOP_GET_DIAGONAL, 199 (void(*)(void))MatGetDiag); CHKERRQ(ierr); 200 201 // Level transfers 202 if (i > 0) { 203 // Interp 204 ierr = PetscMalloc1(1, &userPR[i]); CHKERRQ(ierr); 205 ierr = MatCreateShell(comm, lsize[i], lsize[i-1], gsize[i], gsize[i-1], 206 userPR[i], &matPR[i]); CHKERRQ(ierr); 207 ierr = MatShellSetOperation(matPR[i], MATOP_MULT, 208 (void(*)(void))MatMult_Prolong); 209 CHKERRQ(ierr); 210 ierr = MatShellSetOperation(matPR[i], MATOP_MULT_TRANSPOSE, 211 (void(*)(void))MatMult_Restrict); 212 CHKERRQ(ierr); 213 } 214 } 215 ierr = VecDuplicate(X[fineLevel], &rhs); CHKERRQ(ierr); 216 217 // Set up libCEED 218 CeedInit(ceedresource, &ceed); 219 220 // Print global grid information 221 if (!test_mode) { 222 PetscInt P = degree + 1, Q = P + qextra; 223 const char *usedresource; 224 CeedGetResource(ceed, &usedresource); 225 ierr = PetscPrintf(comm, 226 "\n-- CEED Benchmark Problem %d -- libCEED + PETSc + PCMG --\n" 227 " libCEED:\n" 228 " libCEED Backend : %s\n" 229 " Mesh:\n" 230 " Number of 1D Basis Nodes (p) : %d\n" 231 " Number of 1D Quadrature Points (q) : %d\n" 232 " Global Nodes : %D\n" 233 " Owned Nodes : %D\n" 234 " DoF per node : %D\n" 235 " Multigrid:\n" 236 " Number of Levels : %d\n", 237 bpchoice+1, usedresource, P, Q, 238 gsize[fineLevel]/ncompu, lsize[fineLevel]/ncompu, 239 ncompu, numlevels); CHKERRQ(ierr); 240 } 241 242 // Create RHS vector 243 ierr = VecDuplicate(Xloc[fineLevel], &rhsloc); CHKERRQ(ierr); 244 ierr = VecZeroEntries(rhsloc); CHKERRQ(ierr); 245 ierr = VecGetArray(rhsloc, &r); CHKERRQ(ierr); 246 CeedVectorCreate(ceed, xlsize[fineLevel], &rhsceed); 247 CeedVectorSetArray(rhsceed, CEED_MEM_HOST, CEED_USE_POINTER, r); 248 249 // Set up libCEED operators on each level 250 ierr = PetscMalloc1(numlevels, &ceeddata); CHKERRQ(ierr); 251 for (int i=0; i<numlevels; i++) { 252 // Print level information 253 if (!test_mode && (i == 0 || i == fineLevel)) { 254 ierr = PetscPrintf(comm," Level %D (%s):\n" 255 " Number of 1D Basis Nodes (p) : %d\n" 256 " Global Nodes : %D\n" 257 " Owned Nodes : %D\n", 258 i, (i? "fine" : "coarse"), leveldegrees[i] + 1, 259 gsize[i]/ncompu, lsize[i]/ncompu); CHKERRQ(ierr); 260 } 261 ierr = PetscMalloc1(1, &ceeddata[i]); CHKERRQ(ierr); 262 ierr = SetupLibceedByDegree(dm[i], ceed, leveldegrees[i], dim, qextra, 263 ncompu, gsize[i], xlsize[i], bpchoice, 264 ceeddata[i], i==(fineLevel), rhsceed, 265 &target); CHKERRQ(ierr); 266 } 267 268 // Gather RHS 269 ierr = VecRestoreArray(rhsloc, &r); CHKERRQ(ierr); 270 ierr = VecZeroEntries(rhs); CHKERRQ(ierr); 271 ierr = DMLocalToGlobal(dm[fineLevel], rhsloc, ADD_VALUES, rhs); 272 CHKERRQ(ierr); 273 CeedVectorDestroy(&rhsceed); 274 275 // Create the restriction/interpolation Q-function 276 CeedQFunctionCreateIdentity(ceed, ncompu, CEED_EVAL_NONE, CEED_EVAL_INTERP, 277 &qfrestrict); 278 CeedQFunctionCreateIdentity(ceed, ncompu, CEED_EVAL_INTERP, CEED_EVAL_NONE, 279 &qfprolong); 280 281 // Set up libCEED level transfer operators 282 ierr = CeedLevelTransferSetup(ceed, numlevels, ncompu, bpchoice, ceeddata, 283 leveldegrees, qfrestrict, qfprolong); 284 CHKERRQ(ierr); 285 286 // Create the error Q-function 287 CeedQFunctionCreateInterior(ceed, 1, bpOptions[bpchoice].error, 288 bpOptions[bpchoice].errorfname, &qferror); 289 CeedQFunctionAddInput(qferror, "u", ncompu, CEED_EVAL_INTERP); 290 CeedQFunctionAddInput(qferror, "true_soln", ncompu, CEED_EVAL_NONE); 291 CeedQFunctionAddOutput(qferror, "error", ncompu, CEED_EVAL_NONE); 292 293 // Create the error operator 294 CeedOperatorCreate(ceed, qferror, CEED_QFUNCTION_NONE, CEED_QFUNCTION_NONE, 295 &operror); 296 CeedOperatorSetField(operror, "u", ceeddata[fineLevel]->Erestrictu, 297 ceeddata[fineLevel]->basisu, CEED_VECTOR_ACTIVE); 298 CeedOperatorSetField(operror, "true_soln", ceeddata[fineLevel]->Erestrictui, 299 CEED_BASIS_COLLOCATED, target); 300 CeedOperatorSetField(operror, "error", ceeddata[fineLevel]->Erestrictui, 301 CEED_BASIS_COLLOCATED, CEED_VECTOR_ACTIVE); 302 303 // Calculate multiplicity 304 for (int i=0; i<numlevels; i++) { 305 PetscScalar *x; 306 307 // CEED vector 308 ierr = VecGetArray(Xloc[i], &x); CHKERRQ(ierr); 309 CeedVectorSetArray(ceeddata[i]->xceed, CEED_MEM_HOST, CEED_USE_POINTER, x); 310 311 // Multiplicity 312 CeedElemRestrictionGetMultiplicity(ceeddata[i]->Erestrictu, 313 ceeddata[i]->xceed); 314 315 // Restore vector 316 ierr = VecRestoreArray(Xloc[i], &x); CHKERRQ(ierr); 317 318 // Creat mult vector 319 ierr = VecDuplicate(Xloc[i], &mult[i]); CHKERRQ(ierr); 320 321 // Local-to-global 322 ierr = VecZeroEntries(X[i]); CHKERRQ(ierr); 323 ierr = DMLocalToGlobal(dm[i], Xloc[i], ADD_VALUES, X[i]); 324 CHKERRQ(ierr); 325 ierr = VecZeroEntries(Xloc[i]); CHKERRQ(ierr); 326 327 // Global-to-local 328 ierr = DMGlobalToLocal(dm[i], X[i], INSERT_VALUES, mult[i]); 329 CHKERRQ(ierr); 330 ierr = VecZeroEntries(X[i]); CHKERRQ(ierr); 331 332 // Multiplicity scaling 333 ierr = VecReciprocal(mult[i]); 334 } 335 336 // Set up Mat 337 for (int i=0; i<numlevels; i++) { 338 // User Operator 339 userO[i]->comm = comm; 340 userO[i]->dm = dm[i]; 341 userO[i]->Xloc = Xloc[i]; 342 ierr = VecDuplicate(Xloc[i], &userO[i]->Yloc); CHKERRQ(ierr); 343 userO[i]->xceed = ceeddata[i]->xceed; 344 userO[i]->yceed = ceeddata[i]->yceed; 345 userO[i]->op = ceeddata[i]->opapply; 346 userO[i]->ceed = ceed; 347 348 if (i > 0) { 349 // Prolongation/Restriction Operator 350 userPR[i]->comm = comm; 351 userPR[i]->dmf = dm[i]; 352 userPR[i]->dmc = dm[i-1]; 353 userPR[i]->locvecc = Xloc[i-1]; 354 userPR[i]->locvecf = userO[i]->Yloc; 355 userPR[i]->multvec = mult[i]; 356 userPR[i]->ceedvecc = userO[i-1]->xceed; 357 userPR[i]->ceedvecf = userO[i]->yceed; 358 userPR[i]->opprolong = ceeddata[i]->opprolong; 359 userPR[i]->oprestrict = ceeddata[i]->oprestrict; 360 userPR[i]->ceed = ceed; 361 } 362 } 363 364 // Setup dummy SNES for AMG coarse solve 365 ierr = SNESCreate(comm, &snesdummy); CHKERRQ(ierr); 366 ierr = SNESSetDM(snesdummy, dm[0]); CHKERRQ(ierr); 367 ierr = SNESSetSolution(snesdummy, X[0]); CHKERRQ(ierr); 368 369 // -- Jacobian matrix 370 ierr = DMSetMatType(dm[0], MATAIJ); CHKERRQ(ierr); 371 ierr = DMCreateMatrix(dm[0], &matcoarse); CHKERRQ(ierr); 372 ierr = SNESSetJacobian(snesdummy, matcoarse, matcoarse, NULL, 373 NULL); CHKERRQ(ierr); 374 375 // -- Residual evaluation function 376 ierr = SNESSetFunction(snesdummy, X[0], FormResidual_Ceed, 377 userO[0]); CHKERRQ(ierr); 378 379 // -- Form Jacobian 380 ierr = SNESComputeJacobianDefaultColor(snesdummy, X[0], matO[0], 381 matcoarse, NULL); CHKERRQ(ierr); 382 383 // Set up KSP 384 ierr = KSPCreate(comm, &ksp); CHKERRQ(ierr); 385 { 386 ierr = KSPSetType(ksp, KSPCG); CHKERRQ(ierr); 387 ierr = KSPSetNormType(ksp, KSP_NORM_NATURAL); CHKERRQ(ierr); 388 ierr = KSPSetTolerances(ksp, 1e-10, PETSC_DEFAULT, PETSC_DEFAULT, 389 PETSC_DEFAULT); CHKERRQ(ierr); 390 } 391 ierr = KSPSetFromOptions(ksp); CHKERRQ(ierr); 392 ierr = KSPSetOperators(ksp, matO[fineLevel], matO[fineLevel]); 393 CHKERRQ(ierr); 394 395 // Set up PCMG 396 ierr = KSPGetPC(ksp, &pc); CHKERRQ(ierr); 397 PCMGCycleType pcgmcycletype = PC_MG_CYCLE_V; 398 { 399 ierr = PCSetType(pc, PCMG); CHKERRQ(ierr); 400 401 // PCMG levels 402 ierr = PCMGSetLevels(pc, numlevels, NULL); CHKERRQ(ierr); 403 for (int i=0; i<numlevels; i++) { 404 // Smoother 405 KSP smoother; 406 PC smoother_pc; 407 ierr = PCMGGetSmoother(pc, i, &smoother); CHKERRQ(ierr); 408 ierr = KSPSetType(smoother, KSPCHEBYSHEV); CHKERRQ(ierr); 409 ierr = KSPChebyshevEstEigSet(smoother, 0, 0.1, 0, 1.1); CHKERRQ(ierr); 410 ierr = KSPChebyshevEstEigSetUseNoisy(smoother, PETSC_TRUE); CHKERRQ(ierr); 411 ierr = KSPSetOperators(smoother, matO[i], matO[i]); CHKERRQ(ierr); 412 ierr = KSPGetPC(smoother, &smoother_pc); CHKERRQ(ierr); 413 ierr = PCSetType(smoother_pc, PCJACOBI); CHKERRQ(ierr); 414 ierr = PCJacobiSetType(smoother_pc, PC_JACOBI_DIAGONAL); CHKERRQ(ierr); 415 416 // Work vector 417 if (i < numlevels - 1) { 418 ierr = PCMGSetX(pc, i, X[i]); CHKERRQ(ierr); 419 } 420 421 // Level transfers 422 if (i > 0) { 423 // Interpolation 424 ierr = PCMGSetInterpolation(pc, i, matPR[i]); CHKERRQ(ierr); 425 } 426 427 // Coarse solve 428 KSP coarse; 429 PC coarse_pc; 430 ierr = PCMGGetCoarseSolve(pc, &coarse); CHKERRQ(ierr); 431 ierr = KSPSetType(coarse, KSPPREONLY); CHKERRQ(ierr); 432 ierr = KSPSetOperators(coarse, matcoarse, matcoarse); CHKERRQ(ierr); 433 434 ierr = KSPGetPC(coarse, &coarse_pc); CHKERRQ(ierr); 435 ierr = PCSetType(coarse_pc, PCGAMG); CHKERRQ(ierr); 436 437 ierr = KSPSetOptionsPrefix(coarse, "coarse_"); CHKERRQ(ierr); 438 ierr = PCSetOptionsPrefix(coarse_pc, "coarse_"); CHKERRQ(ierr); 439 ierr = KSPSetFromOptions(coarse); CHKERRQ(ierr); 440 ierr = PCSetFromOptions(coarse_pc); CHKERRQ(ierr); 441 } 442 443 // PCMG options 444 ierr = PCMGSetType(pc, PC_MG_MULTIPLICATIVE); CHKERRQ(ierr); 445 ierr = PCMGSetNumberSmooth(pc, 3); CHKERRQ(ierr); 446 ierr = PCMGSetCycleType(pc, pcgmcycletype); CHKERRQ(ierr); 447 } 448 449 // First run, if benchmarking 450 if (benchmark_mode) { 451 ierr = KSPSetTolerances(ksp, 1e-10, PETSC_DEFAULT, PETSC_DEFAULT, 1); 452 CHKERRQ(ierr); 453 ierr = VecZeroEntries(X[fineLevel]); CHKERRQ(ierr); 454 my_rt_start = MPI_Wtime(); 455 ierr = KSPSolve(ksp, rhs, X[fineLevel]); CHKERRQ(ierr); 456 my_rt = MPI_Wtime() - my_rt_start; 457 ierr = MPI_Allreduce(MPI_IN_PLACE, &my_rt, 1, MPI_DOUBLE, MPI_MIN, comm); 458 CHKERRQ(ierr); 459 // Set maxits based on first iteration timing 460 if (my_rt > 0.02) { 461 ierr = KSPSetTolerances(ksp, 1e-10, PETSC_DEFAULT, PETSC_DEFAULT, 5); 462 CHKERRQ(ierr); 463 } else { 464 ierr = KSPSetTolerances(ksp, 1e-10, PETSC_DEFAULT, PETSC_DEFAULT, 20); 465 CHKERRQ(ierr); 466 } 467 } 468 469 // Timed solve 470 ierr = VecZeroEntries(X[fineLevel]); CHKERRQ(ierr); 471 ierr = PetscBarrier((PetscObject)ksp); CHKERRQ(ierr); 472 my_rt_start = MPI_Wtime(); 473 ierr = KSPSolve(ksp, rhs, X[fineLevel]); CHKERRQ(ierr); 474 my_rt = MPI_Wtime() - my_rt_start; 475 476 // Output results 477 { 478 KSPType ksptype; 479 PCMGType pcmgtype; 480 KSPConvergedReason reason; 481 PetscReal rnorm; 482 PetscInt its; 483 ierr = KSPGetType(ksp, &ksptype); CHKERRQ(ierr); 484 ierr = KSPGetConvergedReason(ksp, &reason); CHKERRQ(ierr); 485 ierr = KSPGetIterationNumber(ksp, &its); CHKERRQ(ierr); 486 ierr = KSPGetResidualNorm(ksp, &rnorm); CHKERRQ(ierr); 487 ierr = PCMGGetType(pc, &pcmgtype); CHKERRQ(ierr); 488 if (!test_mode || reason < 0 || rnorm > 1e-8) { 489 ierr = PetscPrintf(comm, 490 " KSP:\n" 491 " KSP Type : %s\n" 492 " KSP Convergence : %s\n" 493 " Total KSP Iterations : %D\n" 494 " Final rnorm : %e\n", 495 ksptype, KSPConvergedReasons[reason], its, 496 (double)rnorm); CHKERRQ(ierr); 497 ierr = PetscPrintf(comm, 498 " PCMG:\n" 499 " PCMG Type : %s\n" 500 " PCMG Cycle Type : %s\n", 501 PCMGTypes[pcmgtype], 502 PCMGCycleTypes[pcgmcycletype]); CHKERRQ(ierr); 503 } 504 if (!test_mode) { 505 ierr = PetscPrintf(comm," Performance:\n"); CHKERRQ(ierr); 506 } 507 { 508 PetscReal maxerror; 509 ierr = ComputeErrorMax(userO[fineLevel], operror, X[fineLevel], target, 510 &maxerror); CHKERRQ(ierr); 511 PetscReal tol = 5e-2; 512 if (!test_mode || maxerror > tol) { 513 ierr = MPI_Allreduce(&my_rt, &rt_min, 1, MPI_DOUBLE, MPI_MIN, comm); 514 CHKERRQ(ierr); 515 ierr = MPI_Allreduce(&my_rt, &rt_max, 1, MPI_DOUBLE, MPI_MAX, comm); 516 CHKERRQ(ierr); 517 ierr = PetscPrintf(comm, 518 " Pointwise Error (max) : %e\n" 519 " CG Solve Time : %g (%g) sec\n", 520 (double)maxerror, rt_max, rt_min); CHKERRQ(ierr); 521 } 522 } 523 if (benchmark_mode && (!test_mode)) { 524 ierr = PetscPrintf(comm, 525 " DoFs/Sec in CG : %g (%g) million\n", 526 1e-6*gsize[fineLevel]*its/rt_max, 527 1e-6*gsize[fineLevel]*its/rt_min); 528 CHKERRQ(ierr); 529 } 530 } 531 532 if (write_solution) { 533 PetscViewer vtkviewersoln; 534 535 ierr = PetscViewerCreate(comm, &vtkviewersoln); CHKERRQ(ierr); 536 ierr = PetscViewerSetType(vtkviewersoln, PETSCVIEWERVTK); CHKERRQ(ierr); 537 ierr = PetscViewerFileSetName(vtkviewersoln, "solution.vtk"); CHKERRQ(ierr); 538 ierr = VecView(X[fineLevel], vtkviewersoln); CHKERRQ(ierr); 539 ierr = PetscViewerDestroy(&vtkviewersoln); CHKERRQ(ierr); 540 } 541 542 // Cleanup 543 for (int i=0; i<numlevels; i++) { 544 ierr = VecDestroy(&X[i]); CHKERRQ(ierr); 545 ierr = VecDestroy(&Xloc[i]); CHKERRQ(ierr); 546 ierr = VecDestroy(&mult[i]); CHKERRQ(ierr); 547 ierr = VecDestroy(&userO[i]->Yloc); CHKERRQ(ierr); 548 ierr = MatDestroy(&matO[i]); CHKERRQ(ierr); 549 ierr = PetscFree(userO[i]); CHKERRQ(ierr); 550 if (i > 0) { 551 ierr = MatDestroy(&matPR[i]); CHKERRQ(ierr); 552 ierr = PetscFree(userPR[i]); CHKERRQ(ierr); 553 } 554 ierr = CeedDataDestroy(i, ceeddata[i]); CHKERRQ(ierr); 555 ierr = DMDestroy(&dm[i]); CHKERRQ(ierr); 556 } 557 ierr = PetscFree(leveldegrees); CHKERRQ(ierr); 558 ierr = PetscFree(dm); CHKERRQ(ierr); 559 ierr = PetscFree(X); CHKERRQ(ierr); 560 ierr = PetscFree(Xloc); CHKERRQ(ierr); 561 ierr = PetscFree(mult); CHKERRQ(ierr); 562 ierr = PetscFree(matO); CHKERRQ(ierr); 563 ierr = PetscFree(matPR); CHKERRQ(ierr); 564 ierr = PetscFree(ceeddata); CHKERRQ(ierr); 565 ierr = PetscFree(userO); CHKERRQ(ierr); 566 ierr = PetscFree(userPR); CHKERRQ(ierr); 567 ierr = PetscFree(lsize); CHKERRQ(ierr); 568 ierr = PetscFree(xlsize); CHKERRQ(ierr); 569 ierr = PetscFree(gsize); CHKERRQ(ierr); 570 ierr = VecDestroy(&rhs); CHKERRQ(ierr); 571 ierr = VecDestroy(&rhsloc); CHKERRQ(ierr); 572 ierr = MatDestroy(&matcoarse); CHKERRQ(ierr); 573 ierr = KSPDestroy(&ksp); CHKERRQ(ierr); 574 ierr = SNESDestroy(&snesdummy); CHKERRQ(ierr); 575 ierr = DMDestroy(&dmorig); CHKERRQ(ierr); 576 CeedVectorDestroy(&target); 577 CeedQFunctionDestroy(&qferror); 578 CeedQFunctionDestroy(&qfrestrict); 579 CeedQFunctionDestroy(&qfprolong); 580 CeedOperatorDestroy(&operror); 581 CeedDestroy(&ceed); 582 return PetscFinalize(); 583 } 584