1 2 /* 3 Defines matrix-matrix product routines for pairs of MPIAIJ matrices 4 C = A * B 5 */ 6 #include <../src/mat/impls/aij/seq/aij.h> /*I "petscmat.h" I*/ 7 #include <../src/mat/utils/freespace.h> 8 #include <../src/mat/impls/aij/mpi/mpiaij.h> 9 #include <petscbt.h> 10 #include <../src/mat/impls/dense/mpi/mpidense.h> 11 #include <petsc-private/vecimpl.h> 12 13 #undef __FUNCT__ 14 #define __FUNCT__ "MatMatMult_MPIAIJ_MPIAIJ" 15 PetscErrorCode MatMatMult_MPIAIJ_MPIAIJ(Mat A,Mat B,MatReuse scall,PetscReal fill, Mat *C) 16 { 17 PetscErrorCode ierr; 18 19 PetscFunctionBegin; 20 if (scall == MAT_INITIAL_MATRIX) { 21 ierr = MatMatMultSymbolic_MPIAIJ_MPIAIJ(A,B,fill,C);CHKERRQ(ierr); 22 } 23 24 ierr = (*(*C)->ops->matmultnumeric)(A,B,*C);CHKERRQ(ierr); 25 PetscFunctionReturn(0); 26 } 27 28 #undef __FUNCT__ 29 #define __FUNCT__ "MatDestroy_MPIAIJ_MatMatMult" 30 PetscErrorCode MatDestroy_MPIAIJ_MatMatMult(Mat A) 31 { 32 PetscErrorCode ierr; 33 Mat_MPIAIJ *a = (Mat_MPIAIJ*)A->data; 34 Mat_PtAPMPI *ptap = a->ptap; 35 36 PetscFunctionBegin; 37 ierr = PetscFree2(ptap->startsj_s,ptap->startsj_r);CHKERRQ(ierr); 38 ierr = PetscFree(ptap->bufa);CHKERRQ(ierr); 39 ierr = MatDestroy(&ptap->P_loc);CHKERRQ(ierr); 40 ierr = MatDestroy(&ptap->P_oth);CHKERRQ(ierr); 41 ierr = MatDestroy(&ptap->Pt);CHKERRQ(ierr); 42 ierr = PetscFree(ptap->api);CHKERRQ(ierr); 43 ierr = PetscFree(ptap->apj);CHKERRQ(ierr); 44 ierr = PetscFree(ptap->apa);CHKERRQ(ierr); 45 ierr = ptap->destroy(A);CHKERRQ(ierr); 46 ierr = PetscFree(ptap);CHKERRQ(ierr); 47 PetscFunctionReturn(0); 48 } 49 50 #undef __FUNCT__ 51 #define __FUNCT__ "MatDuplicate_MPIAIJ_MatMatMult" 52 PetscErrorCode MatDuplicate_MPIAIJ_MatMatMult(Mat A, MatDuplicateOption op, Mat *M) 53 { 54 PetscErrorCode ierr; 55 Mat_MPIAIJ *a = (Mat_MPIAIJ*)A->data; 56 Mat_PtAPMPI *ptap = a->ptap; 57 58 PetscFunctionBegin; 59 ierr = (*ptap->duplicate)(A,op,M);CHKERRQ(ierr); 60 61 (*M)->ops->destroy = ptap->destroy; /* = MatDestroy_MPIAIJ, *M doesn't duplicate A's special structure! */ 62 (*M)->ops->duplicate = ptap->duplicate; /* = MatDuplicate_MPIAIJ */ 63 PetscFunctionReturn(0); 64 } 65 66 #undef __FUNCT__ 67 #define __FUNCT__ "MatMatMultNumeric_MPIAIJ_MPIAIJ" 68 PetscErrorCode MatMatMultNumeric_MPIAIJ_MPIAIJ(Mat A,Mat P,Mat C) 69 { 70 PetscErrorCode ierr; 71 Mat_MPIAIJ *a =(Mat_MPIAIJ*)A->data,*c=(Mat_MPIAIJ*)C->data; 72 Mat_SeqAIJ *ad =(Mat_SeqAIJ*)(a->A)->data,*ao=(Mat_SeqAIJ*)(a->B)->data; 73 Mat_SeqAIJ *cd =(Mat_SeqAIJ*)(c->A)->data,*co=(Mat_SeqAIJ*)(c->B)->data; 74 PetscInt *adi=ad->i,*adj,*aoi=ao->i,*aoj; 75 PetscScalar *ada,*aoa,*cda=cd->a,*coa=co->a; 76 Mat_SeqAIJ *p_loc,*p_oth; 77 PetscInt *pi_loc,*pj_loc,*pi_oth,*pj_oth,*pj; 78 PetscScalar *pa_loc,*pa_oth,*pa,*apa,valtmp,*ca; 79 PetscInt cm =C->rmap->n,anz,pnz; 80 Mat_PtAPMPI *ptap=c->ptap; 81 PetscInt *api,*apj,*apJ,i,j,k,row; 82 PetscInt cstart=C->cmap->rstart; 83 PetscInt cdnz,conz,k0,k1; 84 85 PetscFunctionBegin; 86 /* 1) get P_oth = ptap->P_oth and P_loc = ptap->P_loc */ 87 /*-----------------------------------------------------*/ 88 /* update numerical values of P_oth and P_loc */ 89 ierr = MatGetBrowsOfAoCols_MPIAIJ(A,P,MAT_REUSE_MATRIX,&ptap->startsj_s,&ptap->startsj_r,&ptap->bufa,&ptap->P_oth);CHKERRQ(ierr); 90 ierr = MatMPIAIJGetLocalMat(P,MAT_REUSE_MATRIX,&ptap->P_loc);CHKERRQ(ierr); 91 92 /* 2) compute numeric C_loc = A_loc*P = Ad*P_loc + Ao*P_oth */ 93 /*----------------------------------------------------------*/ 94 /* get data from symbolic products */ 95 p_loc = (Mat_SeqAIJ*)(ptap->P_loc)->data; 96 p_oth = (Mat_SeqAIJ*)(ptap->P_oth)->data; 97 pi_loc=p_loc->i; pj_loc=p_loc->j; pa_loc=p_loc->a; 98 pi_oth=p_oth->i; pj_oth=p_oth->j; pa_oth=p_oth->a; 99 100 /* get apa for storing dense row A[i,:]*P */ 101 apa = ptap->apa; 102 103 api = ptap->api; 104 apj = ptap->apj; 105 for (i=0; i<cm; i++) { 106 /* diagonal portion of A */ 107 anz = adi[i+1] - adi[i]; 108 adj = ad->j + adi[i]; 109 ada = ad->a + adi[i]; 110 for (j=0; j<anz; j++) { 111 row = adj[j]; 112 pnz = pi_loc[row+1] - pi_loc[row]; 113 pj = pj_loc + pi_loc[row]; 114 pa = pa_loc + pi_loc[row]; 115 116 /* perform dense axpy */ 117 valtmp = ada[j]; 118 for (k=0; k<pnz; k++) { 119 apa[pj[k]] += valtmp*pa[k]; 120 } 121 ierr = PetscLogFlops(2.0*pnz);CHKERRQ(ierr); 122 } 123 124 /* off-diagonal portion of A */ 125 anz = aoi[i+1] - aoi[i]; 126 aoj = ao->j + aoi[i]; 127 aoa = ao->a + aoi[i]; 128 for (j=0; j<anz; j++) { 129 row = aoj[j]; 130 pnz = pi_oth[row+1] - pi_oth[row]; 131 pj = pj_oth + pi_oth[row]; 132 pa = pa_oth + pi_oth[row]; 133 134 /* perform dense axpy */ 135 valtmp = aoa[j]; 136 for (k=0; k<pnz; k++) { 137 apa[pj[k]] += valtmp*pa[k]; 138 } 139 ierr = PetscLogFlops(2.0*pnz);CHKERRQ(ierr); 140 } 141 142 /* set values in C */ 143 apJ = apj + api[i]; 144 cdnz = cd->i[i+1] - cd->i[i]; 145 conz = co->i[i+1] - co->i[i]; 146 147 /* 1st off-diagoanl part of C */ 148 ca = coa + co->i[i]; 149 k = 0; 150 for (k0=0; k0<conz; k0++) { 151 if (apJ[k] >= cstart) break; 152 ca[k0] = apa[apJ[k]]; 153 apa[apJ[k]] = 0.0; 154 k++; 155 } 156 157 /* diagonal part of C */ 158 ca = cda + cd->i[i]; 159 for (k1=0; k1<cdnz; k1++) { 160 ca[k1] = apa[apJ[k]]; 161 apa[apJ[k]] = 0.0; 162 k++; 163 } 164 165 /* 2nd off-diagoanl part of C */ 166 ca = coa + co->i[i]; 167 for (; k0<conz; k0++) { 168 ca[k0] = apa[apJ[k]]; 169 apa[apJ[k]] = 0.0; 170 k++; 171 } 172 } 173 ierr = MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 174 ierr = MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 175 PetscFunctionReturn(0); 176 } 177 178 #undef __FUNCT__ 179 #define __FUNCT__ "MatMatMultSymbolic_MPIAIJ_MPIAIJ" 180 PetscErrorCode MatMatMultSymbolic_MPIAIJ_MPIAIJ(Mat A,Mat P,PetscReal fill,Mat *C) 181 { 182 PetscErrorCode ierr; 183 MPI_Comm comm; 184 Mat Cmpi; 185 Mat_PtAPMPI *ptap; 186 PetscFreeSpaceList free_space=NULL,current_space=NULL; 187 Mat_MPIAIJ *a =(Mat_MPIAIJ*)A->data,*c; 188 Mat_SeqAIJ *ad =(Mat_SeqAIJ*)(a->A)->data,*ao=(Mat_SeqAIJ*)(a->B)->data,*p_loc,*p_oth; 189 PetscInt *pi_loc,*pj_loc,*pi_oth,*pj_oth,*dnz,*onz; 190 PetscInt *adi=ad->i,*adj=ad->j,*aoi=ao->i,*aoj=ao->j,rstart=A->rmap->rstart; 191 PetscInt *lnk,i,pnz,row,*api,*apj,*Jptr,apnz,nspacedouble=0,j,nzi; 192 PetscInt am=A->rmap->n,pN=P->cmap->N,pn=P->cmap->n,pm=P->rmap->n; 193 PetscBT lnkbt; 194 PetscScalar *apa; 195 PetscReal afill; 196 PetscBool scalable=PETSC_TRUE; 197 PetscInt nlnk_max,armax,prmax; 198 199 PetscFunctionBegin; 200 ierr = PetscObjectGetComm((PetscObject)A,&comm);CHKERRQ(ierr); 201 if (A->cmap->rstart != P->rmap->rstart || A->cmap->rend != P->rmap->rend) { 202 SETERRQ4(comm,PETSC_ERR_ARG_SIZ,"Matrix local dimensions are incompatible, (%D, %D) != (%D,%D)",A->cmap->rstart,A->cmap->rend,P->rmap->rstart,P->rmap->rend); 203 } 204 205 ierr = PetscObjectOptionsBegin((PetscObject)A);CHKERRQ(ierr); 206 ierr = PetscOptionsBool("-matmatmult_scalable","Use a scalable but slower C=A*B","",scalable,&scalable,NULL);CHKERRQ(ierr); 207 if (scalable) { 208 ierr = MatMatMultSymbolic_MPIAIJ_MPIAIJ_Scalable(A,P,fill,C);CHKERRQ(ierr); 209 PetscFunctionReturn(0); 210 } 211 ierr = PetscOptionsEnd();CHKERRQ(ierr); 212 213 /* create struct Mat_PtAPMPI and attached it to C later */ 214 ierr = PetscNew(Mat_PtAPMPI,&ptap);CHKERRQ(ierr); 215 216 /* get P_oth by taking rows of P (= non-zero cols of local A) from other processors */ 217 ierr = MatGetBrowsOfAoCols_MPIAIJ(A,P,MAT_INITIAL_MATRIX,&ptap->startsj_s,&ptap->startsj_r,&ptap->bufa,&ptap->P_oth);CHKERRQ(ierr); 218 219 /* get P_loc by taking all local rows of P */ 220 ierr = MatMPIAIJGetLocalMat(P,MAT_INITIAL_MATRIX,&ptap->P_loc);CHKERRQ(ierr); 221 222 p_loc = (Mat_SeqAIJ*)(ptap->P_loc)->data; 223 p_oth = (Mat_SeqAIJ*)(ptap->P_oth)->data; 224 pi_loc = p_loc->i; pj_loc = p_loc->j; 225 pi_oth = p_oth->i; pj_oth = p_oth->j; 226 227 /* first, compute symbolic AP = A_loc*P = A_diag*P_loc + A_off*P_oth */ 228 /*-------------------------------------------------------------------*/ 229 ierr = PetscMalloc((am+2)*sizeof(PetscInt),&api);CHKERRQ(ierr); 230 ptap->api = api; 231 api[0] = 0; 232 233 /* create and initialize a linked list */ 234 armax = ad->rmax+ao->rmax; 235 prmax = PetscMax(p_loc->rmax,p_oth->rmax); 236 nlnk_max = armax*prmax; 237 if (!nlnk_max || nlnk_max > pN) nlnk_max = pN; 238 ierr = PetscLLCondensedCreate(nlnk_max,pN,&lnk,&lnkbt);CHKERRQ(ierr); 239 240 /* Initial FreeSpace size is fill*(nnz(A)+nnz(P)) */ 241 ierr = PetscFreeSpaceGet((PetscInt)(fill*(adi[am]+aoi[am]+pi_loc[pm])),&free_space);CHKERRQ(ierr); 242 243 current_space = free_space; 244 245 ierr = MatPreallocateInitialize(comm,am,pn,dnz,onz);CHKERRQ(ierr); 246 for (i=0; i<am; i++) { 247 /* diagonal portion of A */ 248 nzi = adi[i+1] - adi[i]; 249 for (j=0; j<nzi; j++) { 250 row = *adj++; 251 pnz = pi_loc[row+1] - pi_loc[row]; 252 Jptr = pj_loc + pi_loc[row]; 253 /* add non-zero cols of P into the sorted linked list lnk */ 254 ierr = PetscLLCondensedAddSorted(pnz,Jptr,lnk,lnkbt);CHKERRQ(ierr); 255 } 256 /* off-diagonal portion of A */ 257 nzi = aoi[i+1] - aoi[i]; 258 for (j=0; j<nzi; j++) { 259 row = *aoj++; 260 pnz = pi_oth[row+1] - pi_oth[row]; 261 Jptr = pj_oth + pi_oth[row]; 262 ierr = PetscLLCondensedAddSorted(pnz,Jptr,lnk,lnkbt);CHKERRQ(ierr); 263 } 264 265 apnz = lnk[0]; 266 api[i+1] = api[i] + apnz; 267 268 /* if free space is not available, double the total space in the list */ 269 if (current_space->local_remaining<apnz) { 270 ierr = PetscFreeSpaceGet(apnz+current_space->total_array_size,¤t_space);CHKERRQ(ierr); 271 nspacedouble++; 272 } 273 274 /* Copy data into free space, then initialize lnk */ 275 ierr = PetscLLCondensedClean(pN,apnz,current_space->array,lnk,lnkbt);CHKERRQ(ierr); 276 ierr = MatPreallocateSet(i+rstart,apnz,current_space->array,dnz,onz);CHKERRQ(ierr); 277 278 current_space->array += apnz; 279 current_space->local_used += apnz; 280 current_space->local_remaining -= apnz; 281 } 282 283 /* Allocate space for apj, initialize apj, and */ 284 /* destroy list of free space and other temporary array(s) */ 285 ierr = PetscMalloc((api[am]+1)*sizeof(PetscInt),&ptap->apj);CHKERRQ(ierr); 286 apj = ptap->apj; 287 ierr = PetscFreeSpaceContiguous(&free_space,ptap->apj);CHKERRQ(ierr); 288 ierr = PetscLLDestroy(lnk,lnkbt);CHKERRQ(ierr); 289 290 /* malloc apa to store dense row A[i,:]*P */ 291 ierr = PetscMalloc(pN*sizeof(PetscScalar),&apa);CHKERRQ(ierr); 292 ierr = PetscMemzero(apa,pN*sizeof(PetscScalar));CHKERRQ(ierr); 293 294 ptap->apa = apa; 295 296 /* create and assemble symbolic parallel matrix Cmpi */ 297 /*----------------------------------------------------*/ 298 ierr = MatCreate(comm,&Cmpi);CHKERRQ(ierr); 299 ierr = MatSetSizes(Cmpi,am,pn,PETSC_DETERMINE,PETSC_DETERMINE);CHKERRQ(ierr); 300 ierr = MatSetBlockSizes(Cmpi,A->rmap->bs,P->cmap->bs);CHKERRQ(ierr); 301 302 ierr = MatSetType(Cmpi,MATMPIAIJ);CHKERRQ(ierr); 303 ierr = MatMPIAIJSetPreallocation(Cmpi,0,dnz,0,onz);CHKERRQ(ierr); 304 ierr = MatPreallocateFinalize(dnz,onz);CHKERRQ(ierr); 305 for (i=0; i<am; i++) { 306 row = i + rstart; 307 apnz = api[i+1] - api[i]; 308 ierr = MatSetValues(Cmpi,1,&row,apnz,apj,apa,INSERT_VALUES);CHKERRQ(ierr); 309 apj += apnz; 310 } 311 ierr = MatAssemblyBegin(Cmpi,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 312 ierr = MatAssemblyEnd(Cmpi,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 313 314 ptap->destroy = Cmpi->ops->destroy; 315 ptap->duplicate = Cmpi->ops->duplicate; 316 Cmpi->ops->destroy = MatDestroy_MPIAIJ_MatMatMult; 317 Cmpi->ops->duplicate = MatDuplicate_MPIAIJ_MatMatMult; 318 319 /* attach the supporting struct to Cmpi for reuse */ 320 c = (Mat_MPIAIJ*)Cmpi->data; 321 c->ptap = ptap; 322 323 *C = Cmpi; 324 325 /* set MatInfo */ 326 afill = (PetscReal)api[am]/(adi[am]+aoi[am]+pi_loc[pm]+1) + 1.e-5; 327 if (afill < 1.0) afill = 1.0; 328 Cmpi->info.mallocs = nspacedouble; 329 Cmpi->info.fill_ratio_given = fill; 330 Cmpi->info.fill_ratio_needed = afill; 331 332 #if defined(PETSC_USE_INFO) 333 if (api[am]) { 334 ierr = PetscInfo3(Cmpi,"Reallocs %D; Fill ratio: given %G needed %G.\n",nspacedouble,fill,afill);CHKERRQ(ierr); 335 ierr = PetscInfo1(Cmpi,"Use MatMatMult(A,B,MatReuse,%G,&C) for best performance.;\n",afill);CHKERRQ(ierr); 336 } else { 337 ierr = PetscInfo(Cmpi,"Empty matrix product\n");CHKERRQ(ierr); 338 } 339 #endif 340 PetscFunctionReturn(0); 341 } 342 343 #undef __FUNCT__ 344 #define __FUNCT__ "MatMatMult_MPIAIJ_MPIDense" 345 PetscErrorCode MatMatMult_MPIAIJ_MPIDense(Mat A,Mat B,MatReuse scall,PetscReal fill,Mat *C) 346 { 347 PetscErrorCode ierr; 348 349 PetscFunctionBegin; 350 if (scall == MAT_INITIAL_MATRIX) { 351 ierr = MatMatMultSymbolic_MPIAIJ_MPIDense(A,B,fill,C);CHKERRQ(ierr); 352 } 353 ierr = MatMatMultNumeric_MPIAIJ_MPIDense(A,B,*C);CHKERRQ(ierr); 354 PetscFunctionReturn(0); 355 } 356 357 typedef struct { 358 Mat workB; 359 PetscScalar *rvalues,*svalues; 360 MPI_Request *rwaits,*swaits; 361 } MPIAIJ_MPIDense; 362 363 #undef __FUNCT__ 364 #define __FUNCT__ "MatMPIAIJ_MPIDenseDestroy" 365 PetscErrorCode MatMPIAIJ_MPIDenseDestroy(void *ctx) 366 { 367 MPIAIJ_MPIDense *contents = (MPIAIJ_MPIDense*) ctx; 368 PetscErrorCode ierr; 369 370 PetscFunctionBegin; 371 ierr = MatDestroy(&contents->workB);CHKERRQ(ierr); 372 ierr = PetscFree4(contents->rvalues,contents->svalues,contents->rwaits,contents->swaits);CHKERRQ(ierr); 373 ierr = PetscFree(contents);CHKERRQ(ierr); 374 PetscFunctionReturn(0); 375 } 376 377 #undef __FUNCT__ 378 #define __FUNCT__ "MatMatMultSymbolic_MPIAIJ_MPIDense" 379 PetscErrorCode MatMatMultSymbolic_MPIAIJ_MPIDense(Mat A,Mat B,PetscReal fill,Mat *C) 380 { 381 PetscErrorCode ierr; 382 Mat_MPIAIJ *aij = (Mat_MPIAIJ*) A->data; 383 PetscInt nz = aij->B->cmap->n; 384 PetscContainer container; 385 MPIAIJ_MPIDense *contents; 386 VecScatter ctx = aij->Mvctx; 387 VecScatter_MPI_General *from = (VecScatter_MPI_General*) ctx->fromdata; 388 VecScatter_MPI_General *to = (VecScatter_MPI_General*) ctx->todata; 389 PetscInt m = A->rmap->n,n=B->cmap->n; 390 391 PetscFunctionBegin; 392 ierr = MatCreate(PetscObjectComm((PetscObject)B),C);CHKERRQ(ierr); 393 ierr = MatSetSizes(*C,m,n,A->rmap->N,B->cmap->N);CHKERRQ(ierr); 394 ierr = MatSetBlockSizes(*C,A->rmap->bs,B->cmap->bs);CHKERRQ(ierr); 395 ierr = MatSetType(*C,MATMPIDENSE);CHKERRQ(ierr); 396 ierr = MatMPIDenseSetPreallocation(*C,NULL);CHKERRQ(ierr); 397 ierr = MatAssemblyBegin(*C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 398 ierr = MatAssemblyEnd(*C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 399 400 (*C)->ops->matmultnumeric = MatMatMultNumeric_MPIAIJ_MPIDense; 401 402 ierr = PetscNew(MPIAIJ_MPIDense,&contents);CHKERRQ(ierr); 403 /* Create work matrix used to store off processor rows of B needed for local product */ 404 ierr = MatCreateSeqDense(PETSC_COMM_SELF,nz,B->cmap->N,NULL,&contents->workB);CHKERRQ(ierr); 405 /* Create work arrays needed */ 406 ierr = PetscMalloc4(B->cmap->N*from->starts[from->n],PetscScalar,&contents->rvalues, 407 B->cmap->N*to->starts[to->n],PetscScalar,&contents->svalues, 408 from->n,MPI_Request,&contents->rwaits, 409 to->n,MPI_Request,&contents->swaits);CHKERRQ(ierr); 410 411 ierr = PetscContainerCreate(PetscObjectComm((PetscObject)A),&container);CHKERRQ(ierr); 412 ierr = PetscContainerSetPointer(container,contents);CHKERRQ(ierr); 413 ierr = PetscContainerSetUserDestroy(container,MatMPIAIJ_MPIDenseDestroy);CHKERRQ(ierr); 414 ierr = PetscObjectCompose((PetscObject)(*C),"workB",(PetscObject)container);CHKERRQ(ierr); 415 ierr = PetscContainerDestroy(&container);CHKERRQ(ierr); 416 PetscFunctionReturn(0); 417 } 418 419 #undef __FUNCT__ 420 #define __FUNCT__ "MatMPIDenseScatter" 421 /* 422 Performs an efficient scatter on the rows of B needed by this process; this is 423 a modification of the VecScatterBegin_() routines. 424 */ 425 PetscErrorCode MatMPIDenseScatter(Mat A,Mat B,Mat C,Mat *outworkB) 426 { 427 Mat_MPIAIJ *aij = (Mat_MPIAIJ*)A->data; 428 PetscErrorCode ierr; 429 PetscScalar *b,*w,*svalues,*rvalues; 430 VecScatter ctx = aij->Mvctx; 431 VecScatter_MPI_General *from = (VecScatter_MPI_General*) ctx->fromdata; 432 VecScatter_MPI_General *to = (VecScatter_MPI_General*) ctx->todata; 433 PetscInt i,j,k; 434 PetscInt *sindices,*sstarts,*rindices,*rstarts; 435 PetscMPIInt *sprocs,*rprocs,nrecvs; 436 MPI_Request *swaits,*rwaits; 437 MPI_Comm comm; 438 PetscMPIInt tag = ((PetscObject)ctx)->tag,ncols = B->cmap->N, nrows = aij->B->cmap->n,imdex,nrowsB = B->rmap->n; 439 MPI_Status status; 440 MPIAIJ_MPIDense *contents; 441 PetscContainer container; 442 Mat workB; 443 444 PetscFunctionBegin; 445 ierr = PetscObjectGetComm((PetscObject)A,&comm);CHKERRQ(ierr); 446 ierr = PetscObjectQuery((PetscObject)C,"workB",(PetscObject*)&container);CHKERRQ(ierr); 447 if (!container) SETERRQ(comm,PETSC_ERR_PLIB,"Container does not exist"); 448 ierr = PetscContainerGetPointer(container,(void**)&contents);CHKERRQ(ierr); 449 450 workB = *outworkB = contents->workB; 451 if (nrows != workB->rmap->n) SETERRQ2(comm,PETSC_ERR_PLIB,"Number of rows of workB %D not equal to columns of aij->B %D",nrows,workB->cmap->n); 452 sindices = to->indices; 453 sstarts = to->starts; 454 sprocs = to->procs; 455 swaits = contents->swaits; 456 svalues = contents->svalues; 457 458 rindices = from->indices; 459 rstarts = from->starts; 460 rprocs = from->procs; 461 rwaits = contents->rwaits; 462 rvalues = contents->rvalues; 463 464 ierr = MatDenseGetArray(B,&b);CHKERRQ(ierr); 465 ierr = MatDenseGetArray(workB,&w);CHKERRQ(ierr); 466 467 for (i=0; i<from->n; i++) { 468 ierr = MPI_Irecv(rvalues+ncols*rstarts[i],ncols*(rstarts[i+1]-rstarts[i]),MPIU_SCALAR,rprocs[i],tag,comm,rwaits+i);CHKERRQ(ierr); 469 } 470 471 for (i=0; i<to->n; i++) { 472 /* pack a message at a time */ 473 for (j=0; j<sstarts[i+1]-sstarts[i]; j++) { 474 for (k=0; k<ncols; k++) { 475 svalues[ncols*(sstarts[i] + j) + k] = b[sindices[sstarts[i]+j] + nrowsB*k]; 476 } 477 } 478 ierr = MPI_Isend(svalues+ncols*sstarts[i],ncols*(sstarts[i+1]-sstarts[i]),MPIU_SCALAR,sprocs[i],tag,comm,swaits+i);CHKERRQ(ierr); 479 } 480 481 nrecvs = from->n; 482 while (nrecvs) { 483 ierr = MPI_Waitany(from->n,rwaits,&imdex,&status);CHKERRQ(ierr); 484 nrecvs--; 485 /* unpack a message at a time */ 486 for (j=0; j<rstarts[imdex+1]-rstarts[imdex]; j++) { 487 for (k=0; k<ncols; k++) { 488 w[rindices[rstarts[imdex]+j] + nrows*k] = rvalues[ncols*(rstarts[imdex] + j) + k]; 489 } 490 } 491 } 492 if (to->n) {ierr = MPI_Waitall(to->n,swaits,to->sstatus);CHKERRQ(ierr);} 493 494 ierr = MatDenseRestoreArray(B,&b);CHKERRQ(ierr); 495 ierr = MatDenseRestoreArray(workB,&w);CHKERRQ(ierr); 496 ierr = MatAssemblyBegin(workB,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 497 ierr = MatAssemblyEnd(workB,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 498 PetscFunctionReturn(0); 499 } 500 extern PetscErrorCode MatMatMultNumericAdd_SeqAIJ_SeqDense(Mat,Mat,Mat); 501 502 #undef __FUNCT__ 503 #define __FUNCT__ "MatMatMultNumeric_MPIAIJ_MPIDense" 504 PetscErrorCode MatMatMultNumeric_MPIAIJ_MPIDense(Mat A,Mat B,Mat C) 505 { 506 PetscErrorCode ierr; 507 Mat_MPIAIJ *aij = (Mat_MPIAIJ*)A->data; 508 Mat_MPIDense *bdense = (Mat_MPIDense*)B->data; 509 Mat_MPIDense *cdense = (Mat_MPIDense*)C->data; 510 Mat workB; 511 512 PetscFunctionBegin; 513 /* diagonal block of A times all local rows of B*/ 514 ierr = MatMatMultNumeric_SeqAIJ_SeqDense(aij->A,bdense->A,cdense->A);CHKERRQ(ierr); 515 516 /* get off processor parts of B needed to complete the product */ 517 ierr = MatMPIDenseScatter(A,B,C,&workB);CHKERRQ(ierr); 518 519 /* off-diagonal block of A times nonlocal rows of B */ 520 ierr = MatMatMultNumericAdd_SeqAIJ_SeqDense(aij->B,workB,cdense->A);CHKERRQ(ierr); 521 ierr = MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 522 ierr = MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 523 PetscFunctionReturn(0); 524 } 525 526 #undef __FUNCT__ 527 #define __FUNCT__ "MatMatMultNumeric_MPIAIJ_MPIAIJ_Scalable" 528 PetscErrorCode MatMatMultNumeric_MPIAIJ_MPIAIJ_Scalable(Mat A,Mat P,Mat C) 529 { 530 PetscErrorCode ierr; 531 Mat_MPIAIJ *a = (Mat_MPIAIJ*)A->data,*c=(Mat_MPIAIJ*)C->data; 532 Mat_SeqAIJ *ad = (Mat_SeqAIJ*)(a->A)->data,*ao=(Mat_SeqAIJ*)(a->B)->data; 533 Mat_SeqAIJ *cd = (Mat_SeqAIJ*)(c->A)->data,*co=(Mat_SeqAIJ*)(c->B)->data; 534 PetscInt *adi = ad->i,*adj,*aoi=ao->i,*aoj; 535 PetscScalar *ada,*aoa,*cda=cd->a,*coa=co->a; 536 Mat_SeqAIJ *p_loc,*p_oth; 537 PetscInt *pi_loc,*pj_loc,*pi_oth,*pj_oth,*pj; 538 PetscScalar *pa_loc,*pa_oth,*pa,valtmp,*ca; 539 PetscInt cm = C->rmap->n,anz,pnz; 540 Mat_PtAPMPI *ptap = c->ptap; 541 PetscScalar *apa_sparse = ptap->apa; 542 PetscInt *api,*apj,*apJ,i,j,k,row; 543 PetscInt cstart = C->cmap->rstart; 544 PetscInt cdnz,conz,k0,k1,nextp; 545 546 PetscFunctionBegin; 547 /* 1) get P_oth = ptap->P_oth and P_loc = ptap->P_loc */ 548 /*-----------------------------------------------------*/ 549 /* update numerical values of P_oth and P_loc */ 550 ierr = MatGetBrowsOfAoCols_MPIAIJ(A,P,MAT_REUSE_MATRIX,&ptap->startsj_s,&ptap->startsj_r,&ptap->bufa,&ptap->P_oth);CHKERRQ(ierr); 551 ierr = MatMPIAIJGetLocalMat(P,MAT_REUSE_MATRIX,&ptap->P_loc);CHKERRQ(ierr); 552 553 /* 2) compute numeric C_loc = A_loc*P = Ad*P_loc + Ao*P_oth */ 554 /*----------------------------------------------------------*/ 555 /* get data from symbolic products */ 556 p_loc = (Mat_SeqAIJ*)(ptap->P_loc)->data; 557 p_oth = (Mat_SeqAIJ*)(ptap->P_oth)->data; 558 pi_loc=p_loc->i; pj_loc=p_loc->j; pa_loc=p_loc->a; 559 pi_oth=p_oth->i; pj_oth=p_oth->j; pa_oth=p_oth->a; 560 561 api = ptap->api; 562 apj = ptap->apj; 563 for (i=0; i<cm; i++) { 564 apJ = apj + api[i]; 565 566 /* diagonal portion of A */ 567 anz = adi[i+1] - adi[i]; 568 adj = ad->j + adi[i]; 569 ada = ad->a + adi[i]; 570 for (j=0; j<anz; j++) { 571 row = adj[j]; 572 pnz = pi_loc[row+1] - pi_loc[row]; 573 pj = pj_loc + pi_loc[row]; 574 pa = pa_loc + pi_loc[row]; 575 /* perform sparse axpy */ 576 valtmp = ada[j]; 577 nextp = 0; 578 for (k=0; nextp<pnz; k++) { 579 if (apJ[k] == pj[nextp]) { /* column of AP == column of P */ 580 apa_sparse[k] += valtmp*pa[nextp++]; 581 } 582 } 583 ierr = PetscLogFlops(2.0*pnz);CHKERRQ(ierr); 584 } 585 586 /* off-diagonal portion of A */ 587 anz = aoi[i+1] - aoi[i]; 588 aoj = ao->j + aoi[i]; 589 aoa = ao->a + aoi[i]; 590 for (j=0; j<anz; j++) { 591 row = aoj[j]; 592 pnz = pi_oth[row+1] - pi_oth[row]; 593 pj = pj_oth + pi_oth[row]; 594 pa = pa_oth + pi_oth[row]; 595 /* perform sparse axpy */ 596 valtmp = aoa[j]; 597 nextp = 0; 598 for (k=0; nextp<pnz; k++) { 599 if (apJ[k] == pj[nextp]) { /* column of AP == column of P */ 600 apa_sparse[k] += valtmp*pa[nextp++]; 601 } 602 } 603 ierr = PetscLogFlops(2.0*pnz);CHKERRQ(ierr); 604 } 605 606 /* set values in C */ 607 cdnz = cd->i[i+1] - cd->i[i]; 608 conz = co->i[i+1] - co->i[i]; 609 610 /* 1st off-diagoanl part of C */ 611 ca = coa + co->i[i]; 612 k = 0; 613 for (k0=0; k0<conz; k0++) { 614 if (apJ[k] >= cstart) break; 615 ca[k0] = apa_sparse[k]; 616 apa_sparse[k] = 0.0; 617 k++; 618 } 619 620 /* diagonal part of C */ 621 ca = cda + cd->i[i]; 622 for (k1=0; k1<cdnz; k1++) { 623 ca[k1] = apa_sparse[k]; 624 apa_sparse[k] = 0.0; 625 k++; 626 } 627 628 /* 2nd off-diagoanl part of C */ 629 ca = coa + co->i[i]; 630 for (; k0<conz; k0++) { 631 ca[k0] = apa_sparse[k]; 632 apa_sparse[k] = 0.0; 633 k++; 634 } 635 } 636 ierr = MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 637 ierr = MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 638 PetscFunctionReturn(0); 639 } 640 641 /* same as MatMatMultSymbolic_MPIAIJ_MPIAIJ(), except using LLCondensed to avoid O(BN) memory requirement */ 642 #undef __FUNCT__ 643 #define __FUNCT__ "MatMatMultSymbolic_MPIAIJ_MPIAIJ_Scalable" 644 PetscErrorCode MatMatMultSymbolic_MPIAIJ_MPIAIJ_Scalable(Mat A,Mat P,PetscReal fill,Mat *C) 645 { 646 PetscErrorCode ierr; 647 MPI_Comm comm; 648 Mat Cmpi; 649 Mat_PtAPMPI *ptap; 650 PetscFreeSpaceList free_space = NULL,current_space=NULL; 651 Mat_MPIAIJ *a = (Mat_MPIAIJ*)A->data,*c; 652 Mat_SeqAIJ *ad = (Mat_SeqAIJ*)(a->A)->data,*ao=(Mat_SeqAIJ*)(a->B)->data,*p_loc,*p_oth; 653 PetscInt *pi_loc,*pj_loc,*pi_oth,*pj_oth,*dnz,*onz; 654 PetscInt *adi=ad->i,*adj=ad->j,*aoi=ao->i,*aoj=ao->j,rstart=A->rmap->rstart; 655 PetscInt i,pnz,row,*api,*apj,*Jptr,apnz,nspacedouble=0,j,nzi,*lnk,apnz_max=0; 656 PetscInt am=A->rmap->n,pN=P->cmap->N,pn=P->cmap->n,pm=P->rmap->n; 657 PetscInt nlnk_max,armax,prmax; 658 PetscReal afill; 659 PetscScalar *apa; 660 661 PetscFunctionBegin; 662 ierr = PetscObjectGetComm((PetscObject)A,&comm);CHKERRQ(ierr); 663 /* create struct Mat_PtAPMPI and attached it to C later */ 664 ierr = PetscNew(Mat_PtAPMPI,&ptap);CHKERRQ(ierr); 665 666 /* get P_oth by taking rows of P (= non-zero cols of local A) from other processors */ 667 ierr = MatGetBrowsOfAoCols_MPIAIJ(A,P,MAT_INITIAL_MATRIX,&ptap->startsj_s,&ptap->startsj_r,&ptap->bufa,&ptap->P_oth);CHKERRQ(ierr); 668 669 /* get P_loc by taking all local rows of P */ 670 ierr = MatMPIAIJGetLocalMat(P,MAT_INITIAL_MATRIX,&ptap->P_loc);CHKERRQ(ierr); 671 672 p_loc = (Mat_SeqAIJ*)(ptap->P_loc)->data; 673 p_oth = (Mat_SeqAIJ*)(ptap->P_oth)->data; 674 pi_loc = p_loc->i; pj_loc = p_loc->j; 675 pi_oth = p_oth->i; pj_oth = p_oth->j; 676 677 /* first, compute symbolic AP = A_loc*P = A_diag*P_loc + A_off*P_oth */ 678 /*-------------------------------------------------------------------*/ 679 ierr = PetscMalloc((am+2)*sizeof(PetscInt),&api);CHKERRQ(ierr); 680 ptap->api = api; 681 api[0] = 0; 682 683 /* create and initialize a linked list */ 684 armax = ad->rmax+ao->rmax; 685 prmax = PetscMax(p_loc->rmax,p_oth->rmax); 686 nlnk_max = armax*prmax; 687 if (!nlnk_max || nlnk_max > pN) nlnk_max = pN; 688 ierr = PetscLLCondensedCreate_Scalable(nlnk_max,&lnk);CHKERRQ(ierr); 689 690 /* Initial FreeSpace size is fill*(nnz(A)+nnz(P)) */ 691 ierr = PetscFreeSpaceGet((PetscInt)(fill*(adi[am]+aoi[am]+pi_loc[pm])),&free_space);CHKERRQ(ierr); 692 693 current_space = free_space; 694 695 ierr = MatPreallocateInitialize(comm,am,pn,dnz,onz);CHKERRQ(ierr); 696 for (i=0; i<am; i++) { 697 /* diagonal portion of A */ 698 nzi = adi[i+1] - adi[i]; 699 for (j=0; j<nzi; j++) { 700 row = *adj++; 701 pnz = pi_loc[row+1] - pi_loc[row]; 702 Jptr = pj_loc + pi_loc[row]; 703 /* add non-zero cols of P into the sorted linked list lnk */ 704 ierr = PetscLLCondensedAddSorted_Scalable(pnz,Jptr,lnk);CHKERRQ(ierr); 705 } 706 /* off-diagonal portion of A */ 707 nzi = aoi[i+1] - aoi[i]; 708 for (j=0; j<nzi; j++) { 709 row = *aoj++; 710 pnz = pi_oth[row+1] - pi_oth[row]; 711 Jptr = pj_oth + pi_oth[row]; 712 ierr = PetscLLCondensedAddSorted_Scalable(pnz,Jptr,lnk);CHKERRQ(ierr); 713 } 714 715 apnz = *lnk; 716 api[i+1] = api[i] + apnz; 717 if (apnz > apnz_max) apnz_max = apnz; 718 719 /* if free space is not available, double the total space in the list */ 720 if (current_space->local_remaining<apnz) { 721 ierr = PetscFreeSpaceGet(apnz+current_space->total_array_size,¤t_space);CHKERRQ(ierr); 722 nspacedouble++; 723 } 724 725 /* Copy data into free space, then initialize lnk */ 726 ierr = PetscLLCondensedClean_Scalable(apnz,current_space->array,lnk);CHKERRQ(ierr); 727 ierr = MatPreallocateSet(i+rstart,apnz,current_space->array,dnz,onz);CHKERRQ(ierr); 728 729 current_space->array += apnz; 730 current_space->local_used += apnz; 731 current_space->local_remaining -= apnz; 732 } 733 734 /* Allocate space for apj, initialize apj, and */ 735 /* destroy list of free space and other temporary array(s) */ 736 ierr = PetscMalloc((api[am]+1)*sizeof(PetscInt),&ptap->apj);CHKERRQ(ierr); 737 apj = ptap->apj; 738 ierr = PetscFreeSpaceContiguous(&free_space,ptap->apj);CHKERRQ(ierr); 739 ierr = PetscLLCondensedDestroy_Scalable(lnk);CHKERRQ(ierr); 740 741 /* create and assemble symbolic parallel matrix Cmpi */ 742 /*----------------------------------------------------*/ 743 ierr = MatCreate(comm,&Cmpi);CHKERRQ(ierr); 744 ierr = MatSetSizes(Cmpi,am,pn,PETSC_DETERMINE,PETSC_DETERMINE);CHKERRQ(ierr); 745 ierr = MatSetBlockSizes(Cmpi,A->rmap->bs,P->cmap->bs);CHKERRQ(ierr); 746 ierr = MatSetType(Cmpi,MATMPIAIJ);CHKERRQ(ierr); 747 ierr = MatMPIAIJSetPreallocation(Cmpi,0,dnz,0,onz);CHKERRQ(ierr); 748 ierr = MatPreallocateFinalize(dnz,onz);CHKERRQ(ierr); 749 750 /* malloc apa for assembly Cmpi */ 751 ierr = PetscMalloc(apnz_max*sizeof(PetscScalar),&apa);CHKERRQ(ierr); 752 ierr = PetscMemzero(apa,apnz_max*sizeof(PetscScalar));CHKERRQ(ierr); 753 754 ptap->apa = apa; 755 for (i=0; i<am; i++) { 756 row = i + rstart; 757 apnz = api[i+1] - api[i]; 758 ierr = MatSetValues(Cmpi,1,&row,apnz,apj,apa,INSERT_VALUES);CHKERRQ(ierr); 759 apj += apnz; 760 } 761 ierr = MatAssemblyBegin(Cmpi,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 762 ierr = MatAssemblyEnd(Cmpi,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 763 764 ptap->destroy = Cmpi->ops->destroy; 765 ptap->duplicate = Cmpi->ops->duplicate; 766 Cmpi->ops->matmultnumeric = MatMatMultNumeric_MPIAIJ_MPIAIJ_Scalable; 767 Cmpi->ops->destroy = MatDestroy_MPIAIJ_MatMatMult; 768 Cmpi->ops->duplicate = MatDuplicate_MPIAIJ_MatMatMult; 769 770 /* attach the supporting struct to Cmpi for reuse */ 771 c = (Mat_MPIAIJ*)Cmpi->data; 772 c->ptap = ptap; 773 774 *C = Cmpi; 775 776 /* set MatInfo */ 777 afill = (PetscReal)api[am]/(adi[am]+aoi[am]+pi_loc[pm]+1) + 1.e-5; 778 if (afill < 1.0) afill = 1.0; 779 Cmpi->info.mallocs = nspacedouble; 780 Cmpi->info.fill_ratio_given = fill; 781 Cmpi->info.fill_ratio_needed = afill; 782 783 #if defined(PETSC_USE_INFO) 784 if (api[am]) { 785 ierr = PetscInfo3(Cmpi,"Reallocs %D; Fill ratio: given %G needed %G.\n",nspacedouble,fill,afill);CHKERRQ(ierr); 786 ierr = PetscInfo1(Cmpi,"Use MatMatMult(A,B,MatReuse,%G,&C) for best performance.;\n",afill);CHKERRQ(ierr); 787 } else { 788 ierr = PetscInfo(Cmpi,"Empty matrix product\n");CHKERRQ(ierr); 789 } 790 #endif 791 PetscFunctionReturn(0); 792 } 793 794 /*-------------------------------------------------------------------------*/ 795 #undef __FUNCT__ 796 #define __FUNCT__ "MatTransposeMatMult_MPIAIJ_MPIAIJ" 797 PetscErrorCode MatTransposeMatMult_MPIAIJ_MPIAIJ(Mat P,Mat A,MatReuse scall,PetscReal fill,Mat *C) 798 { 799 PetscErrorCode ierr; 800 PetscBool scalable=PETSC_TRUE,viamatmatmult=PETSC_FALSE; 801 802 PetscFunctionBegin; 803 ierr = PetscOptionsBool("-mattransposematmult_viamatmatmult","Use R=Pt and C=R*A","",viamatmatmult,&viamatmatmult,NULL);CHKERRQ(ierr); 804 if (viamatmatmult) { 805 Mat Pt; 806 Mat_PtAPMPI *ptap; 807 Mat_MPIAIJ *c; 808 if (scall == MAT_INITIAL_MATRIX) { 809 ierr = MatTranspose(P,MAT_INITIAL_MATRIX,&Pt);CHKERRQ(ierr); 810 ierr = MatMatMult(Pt,A,MAT_INITIAL_MATRIX,fill,C);CHKERRQ(ierr); 811 812 c = (Mat_MPIAIJ*)(*C)->data; 813 ptap = c->ptap; 814 ptap->Pt = Pt; 815 } else if (scall == MAT_REUSE_MATRIX) { 816 c = (Mat_MPIAIJ*)(*C)->data; 817 ptap = c->ptap; 818 Pt = ptap->Pt; 819 ierr = MatTranspose(P,scall,&Pt);CHKERRQ(ierr); 820 ierr = MatMatMult(Pt,A,scall,fill,C);CHKERRQ(ierr); 821 } else SETERRQ(PetscObjectComm((PetscObject)A),PETSC_ERR_ARG_WRONGSTATE,"Not supported"); 822 PetscFunctionReturn(0); 823 } 824 825 if (scall == MAT_INITIAL_MATRIX) { 826 ierr = PetscObjectOptionsBegin((PetscObject)A);CHKERRQ(ierr); 827 ierr = PetscOptionsBool("-mattransposematmult_scalable","Use a scalable but slower C=Pt*A","",scalable,&scalable,NULL);CHKERRQ(ierr); 828 if (scalable) { 829 ierr = MatTransposeMatMultSymbolic_MPIAIJ_MPIAIJ_Scalable(P,A,fill,C);CHKERRQ(ierr); 830 } else { 831 ierr = MatTransposeMatMultSymbolic_MPIAIJ_MPIAIJ(P,A,fill,C);CHKERRQ(ierr); 832 } 833 ierr = PetscOptionsEnd();CHKERRQ(ierr); 834 } 835 ierr = (*(*C)->ops->mattransposemultnumeric)(P,A,*C);CHKERRQ(ierr); 836 PetscFunctionReturn(0); 837 } 838 839 #undef __FUNCT__ 840 #define __FUNCT__ "MatTransposeMatMultNumeric_MPIAIJ_MPIAIJ" 841 PetscErrorCode MatTransposeMatMultNumeric_MPIAIJ_MPIAIJ(Mat P,Mat A,Mat C) 842 { 843 PetscErrorCode ierr; 844 Mat_Merge_SeqsToMPI *merge; 845 Mat_MPIAIJ *p =(Mat_MPIAIJ*)P->data,*c=(Mat_MPIAIJ*)C->data; 846 Mat_SeqAIJ *pd=(Mat_SeqAIJ*)(p->A)->data,*po=(Mat_SeqAIJ*)(p->B)->data; 847 Mat_PtAPMPI *ptap; 848 PetscInt *adj,*aJ; 849 PetscInt i,j,k,anz,pnz,row,*cj; 850 MatScalar *ada,*aval,*ca,valtmp; 851 PetscInt am =A->rmap->n,cm=C->rmap->n,pon=(p->B)->cmap->n; 852 MPI_Comm comm; 853 PetscMPIInt size,rank,taga,*len_s; 854 PetscInt *owners,proc,nrows,**buf_ri_k,**nextrow,**nextci; 855 PetscInt **buf_ri,**buf_rj; 856 PetscInt cnz=0,*bj_i,*bi,*bj,bnz,nextcj; /* bi,bj,ba: local array of C(mpi mat) */ 857 MPI_Request *s_waits,*r_waits; 858 MPI_Status *status; 859 MatScalar **abuf_r,*ba_i,*pA,*coa,*ba; 860 PetscInt *ai,*aj,*coi,*coj; 861 PetscInt *poJ,*pdJ; 862 Mat A_loc; 863 Mat_SeqAIJ *a_loc; 864 865 PetscFunctionBegin; 866 ierr = PetscObjectGetComm((PetscObject)C,&comm);CHKERRQ(ierr); 867 ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr); 868 ierr = MPI_Comm_rank(comm,&rank);CHKERRQ(ierr); 869 870 ptap = c->ptap; 871 merge = ptap->merge; 872 873 /* 2) compute numeric C_seq = P_loc^T*A_loc*P - dominating part */ 874 /*--------------------------------------------------------------*/ 875 /* get data from symbolic products */ 876 coi = merge->coi; coj = merge->coj; 877 ierr = PetscMalloc((coi[pon]+1)*sizeof(MatScalar),&coa);CHKERRQ(ierr); 878 ierr = PetscMemzero(coa,coi[pon]*sizeof(MatScalar));CHKERRQ(ierr); 879 880 bi = merge->bi; bj = merge->bj; 881 owners = merge->rowmap->range; 882 ierr = PetscMalloc((bi[cm]+1)*sizeof(MatScalar),&ba);CHKERRQ(ierr); 883 ierr = PetscMemzero(ba,bi[cm]*sizeof(MatScalar));CHKERRQ(ierr); 884 885 /* get A_loc by taking all local rows of A */ 886 A_loc = ptap->A_loc; 887 ierr = MatMPIAIJGetLocalMat(A,MAT_REUSE_MATRIX,&A_loc);CHKERRQ(ierr); 888 a_loc = (Mat_SeqAIJ*)(A_loc)->data; 889 ai = a_loc->i; 890 aj = a_loc->j; 891 892 ierr = PetscMalloc((A->cmap->N)*sizeof(PetscScalar),&aval);CHKERRQ(ierr); /* non-scalable!!! */ 893 ierr = PetscMemzero(aval,A->cmap->N*sizeof(PetscScalar));CHKERRQ(ierr); 894 895 for (i=0; i<am; i++) { 896 /* 2-a) put A[i,:] to dense array aval */ 897 anz = ai[i+1] - ai[i]; 898 adj = aj + ai[i]; 899 ada = a_loc->a + ai[i]; 900 for (j=0; j<anz; j++) { 901 aval[adj[j]] = ada[j]; 902 } 903 904 /* 2-b) Compute Cseq = P_loc[i,:]^T*A[i,:] using outer product */ 905 /*--------------------------------------------------------------*/ 906 /* put the value into Co=(p->B)^T*A (off-diagonal part, send to others) */ 907 pnz = po->i[i+1] - po->i[i]; 908 poJ = po->j + po->i[i]; 909 pA = po->a + po->i[i]; 910 for (j=0; j<pnz; j++) { 911 row = poJ[j]; 912 cnz = coi[row+1] - coi[row]; 913 cj = coj + coi[row]; 914 ca = coa + coi[row]; 915 /* perform dense axpy */ 916 valtmp = pA[j]; 917 for (k=0; k<cnz; k++) { 918 ca[k] += valtmp*aval[cj[k]]; 919 } 920 ierr = PetscLogFlops(2.0*cnz);CHKERRQ(ierr); 921 } 922 923 /* put the value into Cd (diagonal part) */ 924 pnz = pd->i[i+1] - pd->i[i]; 925 pdJ = pd->j + pd->i[i]; 926 pA = pd->a + pd->i[i]; 927 for (j=0; j<pnz; j++) { 928 row = pdJ[j]; 929 cnz = bi[row+1] - bi[row]; 930 cj = bj + bi[row]; 931 ca = ba + bi[row]; 932 /* perform dense axpy */ 933 valtmp = pA[j]; 934 for (k=0; k<cnz; k++) { 935 ca[k] += valtmp*aval[cj[k]]; 936 } 937 ierr = PetscLogFlops(2.0*cnz);CHKERRQ(ierr); 938 } 939 940 /* zero the current row of Pt*A */ 941 aJ = aj + ai[i]; 942 for (k=0; k<anz; k++) aval[aJ[k]] = 0.0; 943 } 944 945 /* 3) send and recv matrix values coa */ 946 /*------------------------------------*/ 947 buf_ri = merge->buf_ri; 948 buf_rj = merge->buf_rj; 949 len_s = merge->len_s; 950 ierr = PetscCommGetNewTag(comm,&taga);CHKERRQ(ierr); 951 ierr = PetscPostIrecvScalar(comm,taga,merge->nrecv,merge->id_r,merge->len_r,&abuf_r,&r_waits);CHKERRQ(ierr); 952 953 ierr = PetscMalloc2(merge->nsend+1,MPI_Request,&s_waits,size,MPI_Status,&status);CHKERRQ(ierr); 954 for (proc=0,k=0; proc<size; proc++) { 955 if (!len_s[proc]) continue; 956 i = merge->owners_co[proc]; 957 ierr = MPI_Isend(coa+coi[i],len_s[proc],MPIU_MATSCALAR,proc,taga,comm,s_waits+k);CHKERRQ(ierr); 958 k++; 959 } 960 if (merge->nrecv) {ierr = MPI_Waitall(merge->nrecv,r_waits,status);CHKERRQ(ierr);} 961 if (merge->nsend) {ierr = MPI_Waitall(merge->nsend,s_waits,status);CHKERRQ(ierr);} 962 963 ierr = PetscFree2(s_waits,status);CHKERRQ(ierr); 964 ierr = PetscFree(r_waits);CHKERRQ(ierr); 965 ierr = PetscFree(coa);CHKERRQ(ierr); 966 967 /* 4) insert local Cseq and received values into Cmpi */ 968 /*----------------------------------------------------*/ 969 ierr = PetscMalloc3(merge->nrecv,PetscInt**,&buf_ri_k,merge->nrecv,PetscInt*,&nextrow,merge->nrecv,PetscInt*,&nextci);CHKERRQ(ierr); 970 for (k=0; k<merge->nrecv; k++) { 971 buf_ri_k[k] = buf_ri[k]; /* beginning of k-th recved i-structure */ 972 nrows = *(buf_ri_k[k]); 973 nextrow[k] = buf_ri_k[k]+1; /* next row number of k-th recved i-structure */ 974 nextci[k] = buf_ri_k[k] + (nrows + 1); /* poins to the next i-structure of k-th recved i-structure */ 975 } 976 977 for (i=0; i<cm; i++) { 978 row = owners[rank] + i; /* global row index of C_seq */ 979 bj_i = bj + bi[i]; /* col indices of the i-th row of C */ 980 ba_i = ba + bi[i]; 981 bnz = bi[i+1] - bi[i]; 982 /* add received vals into ba */ 983 for (k=0; k<merge->nrecv; k++) { /* k-th received message */ 984 /* i-th row */ 985 if (i == *nextrow[k]) { 986 cnz = *(nextci[k]+1) - *nextci[k]; 987 cj = buf_rj[k] + *(nextci[k]); 988 ca = abuf_r[k] + *(nextci[k]); 989 nextcj = 0; 990 for (j=0; nextcj<cnz; j++) { 991 if (bj_i[j] == cj[nextcj]) { /* bcol == ccol */ 992 ba_i[j] += ca[nextcj++]; 993 } 994 } 995 nextrow[k]++; nextci[k]++; 996 ierr = PetscLogFlops(2.0*cnz);CHKERRQ(ierr); 997 } 998 } 999 ierr = MatSetValues(C,1,&row,bnz,bj_i,ba_i,INSERT_VALUES);CHKERRQ(ierr); 1000 } 1001 ierr = MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1002 ierr = MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1003 1004 ierr = PetscFree(ba);CHKERRQ(ierr); 1005 ierr = PetscFree(abuf_r[0]);CHKERRQ(ierr); 1006 ierr = PetscFree(abuf_r);CHKERRQ(ierr); 1007 ierr = PetscFree3(buf_ri_k,nextrow,nextci);CHKERRQ(ierr); 1008 ierr = PetscFree(aval);CHKERRQ(ierr); 1009 PetscFunctionReturn(0); 1010 } 1011 1012 /* This routine is modified from MatPtAPSymbolic_MPIAIJ_MPIAIJ() */ 1013 #undef __FUNCT__ 1014 #define __FUNCT__ "MatTransposeMatMultSymbolic_MPIAIJ_MPIAIJ" 1015 PetscErrorCode MatTransposeMatMultSymbolic_MPIAIJ_MPIAIJ(Mat P,Mat A,PetscReal fill,Mat *C) 1016 { 1017 PetscErrorCode ierr; 1018 Mat Cmpi,A_loc,POt,PDt; 1019 Mat_PtAPMPI *ptap; 1020 PetscFreeSpaceList free_space=NULL,current_space=NULL; 1021 Mat_MPIAIJ *p =(Mat_MPIAIJ*)P->data,*c; 1022 PetscInt *pdti,*pdtj,*poti,*potj,*ptJ; 1023 PetscInt nnz; 1024 PetscInt *lnk,*owners_co,*coi,*coj,i,k,pnz,row; 1025 PetscInt am=A->rmap->n,pn=P->cmap->n; 1026 PetscBT lnkbt; 1027 MPI_Comm comm; 1028 PetscMPIInt size,rank,tagi,tagj,*len_si,*len_s,*len_ri; 1029 PetscInt **buf_rj,**buf_ri,**buf_ri_k; 1030 PetscInt len,proc,*dnz,*onz,*owners; 1031 PetscInt nzi,*bi,*bj; 1032 PetscInt nrows,*buf_s,*buf_si,*buf_si_i,**nextrow,**nextci; 1033 MPI_Request *swaits,*rwaits; 1034 MPI_Status *sstatus,rstatus; 1035 Mat_Merge_SeqsToMPI *merge; 1036 PetscInt *ai,*aj,*Jptr,anz,*prmap=p->garray,pon,nspacedouble=0,j; 1037 PetscReal afill =1.0,afill_tmp; 1038 PetscInt rstart = P->cmap->rstart,rmax,aN=A->cmap->N,Crmax; 1039 PetscScalar *vals; 1040 Mat_SeqAIJ *a_loc, *pdt,*pot; 1041 1042 PetscFunctionBegin; 1043 ierr = PetscObjectGetComm((PetscObject)A,&comm);CHKERRQ(ierr); 1044 /* check if matrix local sizes are compatible */ 1045 if (A->rmap->rstart != P->rmap->rstart || A->rmap->rend != P->rmap->rend) { 1046 SETERRQ4(comm,PETSC_ERR_ARG_SIZ,"Matrix local dimensions are incompatible, A (%D, %D) != P (%D,%D)",A->rmap->rstart,A->rmap->rend,P->rmap->rstart,P->rmap->rend); 1047 } 1048 1049 ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr); 1050 ierr = MPI_Comm_rank(comm,&rank);CHKERRQ(ierr); 1051 1052 /* create struct Mat_PtAPMPI and attached it to C later */ 1053 ierr = PetscNew(Mat_PtAPMPI,&ptap);CHKERRQ(ierr); 1054 1055 /* get A_loc by taking all local rows of A */ 1056 ierr = MatMPIAIJGetLocalMat(A,MAT_INITIAL_MATRIX,&A_loc);CHKERRQ(ierr); 1057 1058 ptap->A_loc = A_loc; 1059 1060 a_loc = (Mat_SeqAIJ*)(A_loc)->data; 1061 ai = a_loc->i; 1062 aj = a_loc->j; 1063 1064 /* determine symbolic Co=(p->B)^T*A - send to others */ 1065 /*----------------------------------------------------*/ 1066 ierr = MatTransposeSymbolic_SeqAIJ(p->A,&PDt);CHKERRQ(ierr); 1067 pdt = (Mat_SeqAIJ*)PDt->data; 1068 pdti = pdt->i; pdtj = pdt->j; 1069 1070 ierr = MatTransposeSymbolic_SeqAIJ(p->B,&POt);CHKERRQ(ierr); 1071 pot = (Mat_SeqAIJ*)POt->data; 1072 poti = pot->i; potj = pot->j; 1073 1074 /* then, compute symbolic Co = (p->B)^T*A */ 1075 pon = (p->B)->cmap->n; /* total num of rows to be sent to other processors >= (num of nonzero rows of C_seq) - pn */ 1076 ierr = PetscMalloc((pon+1)*sizeof(PetscInt),&coi);CHKERRQ(ierr); 1077 coi[0] = 0; 1078 1079 /* set initial free space to be fill*(nnz(p->B) + nnz(A)) */ 1080 nnz = fill*(poti[pon] + ai[am]); 1081 ierr = PetscFreeSpaceGet(nnz,&free_space);CHKERRQ(ierr); 1082 current_space = free_space; 1083 1084 /* create and initialize a linked list */ 1085 i = PetscMax(pdt->rmax,pot->rmax); 1086 Crmax = i*a_loc->rmax*size; 1087 if (!Crmax || Crmax > aN) Crmax = aN; 1088 ierr = PetscLLCondensedCreate(Crmax,aN,&lnk,&lnkbt);CHKERRQ(ierr); 1089 1090 for (i=0; i<pon; i++) { 1091 pnz = poti[i+1] - poti[i]; 1092 ptJ = potj + poti[i]; 1093 for (j=0; j<pnz; j++) { 1094 row = ptJ[j]; /* row of A_loc == col of Pot */ 1095 anz = ai[row+1] - ai[row]; 1096 Jptr = aj + ai[row]; 1097 /* add non-zero cols of AP into the sorted linked list lnk */ 1098 ierr = PetscLLCondensedAddSorted(anz,Jptr,lnk,lnkbt);CHKERRQ(ierr); 1099 } 1100 nnz = lnk[0]; 1101 1102 /* If free space is not available, double the total space in the list */ 1103 if (current_space->local_remaining<nnz) { 1104 ierr = PetscFreeSpaceGet(nnz+current_space->total_array_size,¤t_space);CHKERRQ(ierr); 1105 nspacedouble++; 1106 } 1107 1108 /* Copy data into free space, and zero out denserows */ 1109 ierr = PetscLLCondensedClean(aN,nnz,current_space->array,lnk,lnkbt);CHKERRQ(ierr); 1110 1111 current_space->array += nnz; 1112 current_space->local_used += nnz; 1113 current_space->local_remaining -= nnz; 1114 1115 coi[i+1] = coi[i] + nnz; 1116 } 1117 1118 ierr = PetscMalloc((coi[pon]+1)*sizeof(PetscInt),&coj);CHKERRQ(ierr); 1119 ierr = PetscFreeSpaceContiguous(&free_space,coj);CHKERRQ(ierr); 1120 1121 afill_tmp = (PetscReal)coi[pon]/(poti[pon] + ai[am]+1); 1122 if (afill_tmp > afill) afill = afill_tmp; 1123 1124 /* send j-array (coj) of Co to other processors */ 1125 /*----------------------------------------------*/ 1126 /* determine row ownership */ 1127 ierr = PetscNew(Mat_Merge_SeqsToMPI,&merge);CHKERRQ(ierr); 1128 ierr = PetscLayoutCreate(comm,&merge->rowmap);CHKERRQ(ierr); 1129 1130 merge->rowmap->n = pn; 1131 merge->rowmap->bs = 1; 1132 1133 ierr = PetscLayoutSetUp(merge->rowmap);CHKERRQ(ierr); 1134 owners = merge->rowmap->range; 1135 1136 /* determine the number of messages to send, their lengths */ 1137 ierr = PetscMalloc(size*sizeof(PetscMPIInt),&len_si);CHKERRQ(ierr); 1138 ierr = PetscMemzero(len_si,size*sizeof(PetscMPIInt));CHKERRQ(ierr); 1139 ierr = PetscMalloc(size*sizeof(PetscMPIInt),&merge->len_s);CHKERRQ(ierr); 1140 1141 len_s = merge->len_s; 1142 merge->nsend = 0; 1143 1144 ierr = PetscMalloc((size+2)*sizeof(PetscInt),&owners_co);CHKERRQ(ierr); 1145 ierr = PetscMemzero(len_s,size*sizeof(PetscMPIInt));CHKERRQ(ierr); 1146 1147 proc = 0; 1148 for (i=0; i<pon; i++) { 1149 while (prmap[i] >= owners[proc+1]) proc++; 1150 len_si[proc]++; /* num of rows in Co to be sent to [proc] */ 1151 len_s[proc] += coi[i+1] - coi[i]; 1152 } 1153 1154 len = 0; /* max length of buf_si[] */ 1155 owners_co[0] = 0; 1156 for (proc=0; proc<size; proc++) { 1157 owners_co[proc+1] = owners_co[proc] + len_si[proc]; 1158 if (len_si[proc]) { 1159 merge->nsend++; 1160 len_si[proc] = 2*(len_si[proc] + 1); 1161 len += len_si[proc]; 1162 } 1163 } 1164 1165 /* determine the number and length of messages to receive for coi and coj */ 1166 ierr = PetscGatherNumberOfMessages(comm,NULL,len_s,&merge->nrecv);CHKERRQ(ierr); 1167 ierr = PetscGatherMessageLengths2(comm,merge->nsend,merge->nrecv,len_s,len_si,&merge->id_r,&merge->len_r,&len_ri);CHKERRQ(ierr); 1168 1169 /* post the Irecv and Isend of coj */ 1170 ierr = PetscCommGetNewTag(comm,&tagj);CHKERRQ(ierr); 1171 ierr = PetscPostIrecvInt(comm,tagj,merge->nrecv,merge->id_r,merge->len_r,&buf_rj,&rwaits);CHKERRQ(ierr); 1172 ierr = PetscMalloc((merge->nsend+1)*sizeof(MPI_Request),&swaits);CHKERRQ(ierr); 1173 for (proc=0, k=0; proc<size; proc++) { 1174 if (!len_s[proc]) continue; 1175 i = owners_co[proc]; 1176 ierr = MPI_Isend(coj+coi[i],len_s[proc],MPIU_INT,proc,tagj,comm,swaits+k);CHKERRQ(ierr); 1177 k++; 1178 } 1179 1180 /* receives and sends of coj are complete */ 1181 ierr = PetscMalloc(size*sizeof(MPI_Status),&sstatus);CHKERRQ(ierr); 1182 for (i=0; i<merge->nrecv; i++) { 1183 PetscMPIInt icompleted; 1184 ierr = MPI_Waitany(merge->nrecv,rwaits,&icompleted,&rstatus);CHKERRQ(ierr); 1185 } 1186 ierr = PetscFree(rwaits);CHKERRQ(ierr); 1187 if (merge->nsend) {ierr = MPI_Waitall(merge->nsend,swaits,sstatus);CHKERRQ(ierr);} 1188 1189 /* send and recv coi */ 1190 /*-------------------*/ 1191 ierr = PetscCommGetNewTag(comm,&tagi);CHKERRQ(ierr); 1192 ierr = PetscPostIrecvInt(comm,tagi,merge->nrecv,merge->id_r,len_ri,&buf_ri,&rwaits);CHKERRQ(ierr); 1193 ierr = PetscMalloc((len+1)*sizeof(PetscInt),&buf_s);CHKERRQ(ierr); 1194 buf_si = buf_s; /* points to the beginning of k-th msg to be sent */ 1195 for (proc=0,k=0; proc<size; proc++) { 1196 if (!len_s[proc]) continue; 1197 /* form outgoing message for i-structure: 1198 buf_si[0]: nrows to be sent 1199 [1:nrows]: row index (global) 1200 [nrows+1:2*nrows+1]: i-structure index 1201 */ 1202 /*-------------------------------------------*/ 1203 nrows = len_si[proc]/2 - 1; 1204 buf_si_i = buf_si + nrows+1; 1205 buf_si[0] = nrows; 1206 buf_si_i[0] = 0; 1207 nrows = 0; 1208 for (i=owners_co[proc]; i<owners_co[proc+1]; i++) { 1209 nzi = coi[i+1] - coi[i]; 1210 buf_si_i[nrows+1] = buf_si_i[nrows] + nzi; /* i-structure */ 1211 buf_si[nrows+1] = prmap[i] -owners[proc]; /* local row index */ 1212 nrows++; 1213 } 1214 ierr = MPI_Isend(buf_si,len_si[proc],MPIU_INT,proc,tagi,comm,swaits+k);CHKERRQ(ierr); 1215 k++; 1216 buf_si += len_si[proc]; 1217 } 1218 i = merge->nrecv; 1219 while (i--) { 1220 PetscMPIInt icompleted; 1221 ierr = MPI_Waitany(merge->nrecv,rwaits,&icompleted,&rstatus);CHKERRQ(ierr); 1222 } 1223 ierr = PetscFree(rwaits);CHKERRQ(ierr); 1224 if (merge->nsend) {ierr = MPI_Waitall(merge->nsend,swaits,sstatus);CHKERRQ(ierr);} 1225 ierr = PetscFree(len_si);CHKERRQ(ierr); 1226 ierr = PetscFree(len_ri);CHKERRQ(ierr); 1227 ierr = PetscFree(swaits);CHKERRQ(ierr); 1228 ierr = PetscFree(sstatus);CHKERRQ(ierr); 1229 ierr = PetscFree(buf_s);CHKERRQ(ierr); 1230 1231 /* compute the local portion of C (mpi mat) */ 1232 /*------------------------------------------*/ 1233 /* allocate bi array and free space for accumulating nonzero column info */ 1234 ierr = PetscMalloc((pn+1)*sizeof(PetscInt),&bi);CHKERRQ(ierr); 1235 bi[0] = 0; 1236 1237 /* set initial free space to be fill*(nnz(P) + nnz(A)) */ 1238 nnz = fill*(pdti[pn] + poti[pon] + ai[am]); 1239 ierr = PetscFreeSpaceGet(nnz,&free_space);CHKERRQ(ierr); 1240 current_space = free_space; 1241 1242 ierr = PetscMalloc3(merge->nrecv,PetscInt**,&buf_ri_k,merge->nrecv,PetscInt*,&nextrow,merge->nrecv,PetscInt*,&nextci);CHKERRQ(ierr); 1243 for (k=0; k<merge->nrecv; k++) { 1244 buf_ri_k[k] = buf_ri[k]; /* beginning of k-th recved i-structure */ 1245 nrows = *buf_ri_k[k]; 1246 nextrow[k] = buf_ri_k[k] + 1; /* next row number of k-th recved i-structure */ 1247 nextci[k] = buf_ri_k[k] + (nrows + 1); /* poins to the next i-structure of k-th recved i-structure */ 1248 } 1249 1250 ierr = MatPreallocateInitialize(comm,pn,A->cmap->n,dnz,onz);CHKERRQ(ierr); 1251 rmax = 0; 1252 for (i=0; i<pn; i++) { 1253 /* add pdt[i,:]*AP into lnk */ 1254 pnz = pdti[i+1] - pdti[i]; 1255 ptJ = pdtj + pdti[i]; 1256 for (j=0; j<pnz; j++) { 1257 row = ptJ[j]; /* row of AP == col of Pt */ 1258 anz = ai[row+1] - ai[row]; 1259 Jptr = aj + ai[row]; 1260 /* add non-zero cols of AP into the sorted linked list lnk */ 1261 ierr = PetscLLCondensedAddSorted(anz,Jptr,lnk,lnkbt);CHKERRQ(ierr); 1262 } 1263 1264 /* add received col data into lnk */ 1265 for (k=0; k<merge->nrecv; k++) { /* k-th received message */ 1266 if (i == *nextrow[k]) { /* i-th row */ 1267 nzi = *(nextci[k]+1) - *nextci[k]; 1268 Jptr = buf_rj[k] + *nextci[k]; 1269 ierr = PetscLLCondensedAddSorted(nzi,Jptr,lnk,lnkbt);CHKERRQ(ierr); 1270 nextrow[k]++; nextci[k]++; 1271 } 1272 } 1273 nnz = lnk[0]; 1274 1275 /* if free space is not available, make more free space */ 1276 if (current_space->local_remaining<nnz) { 1277 ierr = PetscFreeSpaceGet(nnz+current_space->total_array_size,¤t_space);CHKERRQ(ierr); 1278 nspacedouble++; 1279 } 1280 /* copy data into free space, then initialize lnk */ 1281 ierr = PetscLLCondensedClean(aN,nnz,current_space->array,lnk,lnkbt);CHKERRQ(ierr); 1282 ierr = MatPreallocateSet(i+owners[rank],nnz,current_space->array,dnz,onz);CHKERRQ(ierr); 1283 1284 current_space->array += nnz; 1285 current_space->local_used += nnz; 1286 current_space->local_remaining -= nnz; 1287 1288 bi[i+1] = bi[i] + nnz; 1289 if (nnz > rmax) rmax = nnz; 1290 } 1291 ierr = PetscFree3(buf_ri_k,nextrow,nextci);CHKERRQ(ierr); 1292 1293 ierr = PetscMalloc((bi[pn]+1)*sizeof(PetscInt),&bj);CHKERRQ(ierr); 1294 ierr = PetscFreeSpaceContiguous(&free_space,bj);CHKERRQ(ierr); 1295 1296 afill_tmp = (PetscReal)bi[pn]/(pdti[pn] + poti[pon] + ai[am]+1); 1297 if (afill_tmp > afill) afill = afill_tmp; 1298 ierr = PetscLLCondensedDestroy(lnk,lnkbt);CHKERRQ(ierr); 1299 ierr = MatDestroy(&POt);CHKERRQ(ierr); 1300 ierr = MatDestroy(&PDt);CHKERRQ(ierr); 1301 1302 /* create symbolic parallel matrix Cmpi - why cannot be assembled in Numeric part */ 1303 /*----------------------------------------------------------------------------------*/ 1304 ierr = PetscMalloc((rmax+1)*sizeof(PetscScalar),&vals);CHKERRQ(ierr); 1305 ierr = PetscMemzero(vals,rmax*sizeof(PetscScalar));CHKERRQ(ierr); 1306 1307 ierr = MatCreate(comm,&Cmpi);CHKERRQ(ierr); 1308 ierr = MatSetSizes(Cmpi,pn,A->cmap->n,PETSC_DETERMINE,PETSC_DETERMINE);CHKERRQ(ierr); 1309 ierr = MatSetBlockSizes(Cmpi,P->cmap->bs,A->cmap->bs);CHKERRQ(ierr); 1310 ierr = MatSetType(Cmpi,MATMPIAIJ);CHKERRQ(ierr); 1311 ierr = MatMPIAIJSetPreallocation(Cmpi,0,dnz,0,onz);CHKERRQ(ierr); 1312 ierr = MatPreallocateFinalize(dnz,onz);CHKERRQ(ierr); 1313 ierr = MatSetBlockSize(Cmpi,1);CHKERRQ(ierr); 1314 for (i=0; i<pn; i++) { 1315 row = i + rstart; 1316 nnz = bi[i+1] - bi[i]; 1317 Jptr = bj + bi[i]; 1318 ierr = MatSetValues(Cmpi,1,&row,nnz,Jptr,vals,INSERT_VALUES);CHKERRQ(ierr); 1319 } 1320 ierr = MatAssemblyBegin(Cmpi,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1321 ierr = MatAssemblyEnd(Cmpi,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1322 ierr = PetscFree(vals);CHKERRQ(ierr); 1323 1324 merge->bi = bi; 1325 merge->bj = bj; 1326 merge->coi = coi; 1327 merge->coj = coj; 1328 merge->buf_ri = buf_ri; 1329 merge->buf_rj = buf_rj; 1330 merge->owners_co = owners_co; 1331 merge->destroy = Cmpi->ops->destroy; 1332 merge->duplicate = Cmpi->ops->duplicate; 1333 1334 Cmpi->ops->mattransposemultnumeric = MatTransposeMatMultNumeric_MPIAIJ_MPIAIJ; 1335 Cmpi->ops->destroy = MatDestroy_MPIAIJ_PtAP; 1336 1337 /* attach the supporting struct to Cmpi for reuse */ 1338 c = (Mat_MPIAIJ*)Cmpi->data; 1339 c->ptap = ptap; 1340 ptap->api = NULL; 1341 ptap->apj = NULL; 1342 ptap->merge = merge; 1343 ptap->rmax = rmax; 1344 1345 *C = Cmpi; 1346 #if defined(PETSC_USE_INFO) 1347 if (bi[pn] != 0) { 1348 ierr = PetscInfo3(Cmpi,"Reallocs %D; Fill ratio: given %G needed %G.\n",nspacedouble,fill,afill);CHKERRQ(ierr); 1349 ierr = PetscInfo1(Cmpi,"Use MatTransposeMatMult(A,B,MatReuse,%G,&C) for best performance.\n",afill);CHKERRQ(ierr); 1350 } else { 1351 ierr = PetscInfo(Cmpi,"Empty matrix product\n");CHKERRQ(ierr); 1352 } 1353 #endif 1354 PetscFunctionReturn(0); 1355 } 1356 1357 #undef __FUNCT__ 1358 #define __FUNCT__ "MatTransposeMatMultNumeric_MPIAIJ_MPIAIJ_Scalable" 1359 PetscErrorCode MatTransposeMatMultNumeric_MPIAIJ_MPIAIJ_Scalable(Mat P,Mat A,Mat C) 1360 { 1361 PetscErrorCode ierr; 1362 Mat_Merge_SeqsToMPI *merge; 1363 Mat_MPIAIJ *p =(Mat_MPIAIJ*)P->data,*c=(Mat_MPIAIJ*)C->data; 1364 Mat_SeqAIJ *pd=(Mat_SeqAIJ*)(p->A)->data,*po=(Mat_SeqAIJ*)(p->B)->data; 1365 Mat_PtAPMPI *ptap; 1366 PetscInt *adj; 1367 PetscInt i,j,k,anz,pnz,row,*cj,nexta; 1368 MatScalar *ada,*ca,valtmp; 1369 PetscInt am =A->rmap->n,cm=C->rmap->n,pon=(p->B)->cmap->n; 1370 MPI_Comm comm; 1371 PetscMPIInt size,rank,taga,*len_s; 1372 PetscInt *owners,proc,nrows,**buf_ri_k,**nextrow,**nextci; 1373 PetscInt **buf_ri,**buf_rj; 1374 PetscInt cnz=0,*bj_i,*bi,*bj,bnz,nextcj; /* bi,bj,ba: local array of C(mpi mat) */ 1375 MPI_Request *s_waits,*r_waits; 1376 MPI_Status *status; 1377 MatScalar **abuf_r,*ba_i,*pA,*coa,*ba; 1378 PetscInt *ai,*aj,*coi,*coj; 1379 PetscInt *poJ,*pdJ; 1380 Mat A_loc; 1381 Mat_SeqAIJ *a_loc; 1382 1383 PetscFunctionBegin; 1384 ierr = PetscObjectGetComm((PetscObject)C,&comm);CHKERRQ(ierr); 1385 ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr); 1386 ierr = MPI_Comm_rank(comm,&rank);CHKERRQ(ierr); 1387 1388 ptap = c->ptap; 1389 merge = ptap->merge; 1390 1391 /* 2) compute numeric C_seq = P_loc^T*A_loc */ 1392 /*------------------------------------------*/ 1393 /* get data from symbolic products */ 1394 coi = merge->coi; coj = merge->coj; 1395 ierr = PetscMalloc((coi[pon]+1)*sizeof(MatScalar),&coa);CHKERRQ(ierr); 1396 ierr = PetscMemzero(coa,coi[pon]*sizeof(MatScalar));CHKERRQ(ierr); 1397 bi = merge->bi; bj = merge->bj; 1398 owners = merge->rowmap->range; 1399 ierr = PetscMalloc((bi[cm]+1)*sizeof(MatScalar),&ba);CHKERRQ(ierr); 1400 ierr = PetscMemzero(ba,bi[cm]*sizeof(MatScalar));CHKERRQ(ierr); 1401 1402 /* get A_loc by taking all local rows of A */ 1403 A_loc = ptap->A_loc; 1404 ierr = MatMPIAIJGetLocalMat(A,MAT_REUSE_MATRIX,&A_loc);CHKERRQ(ierr); 1405 a_loc = (Mat_SeqAIJ*)(A_loc)->data; 1406 ai = a_loc->i; 1407 aj = a_loc->j; 1408 1409 for (i=0; i<am; i++) { 1410 anz = ai[i+1] - ai[i]; 1411 adj = aj + ai[i]; 1412 ada = a_loc->a + ai[i]; 1413 1414 /* 2-b) Compute Cseq = P_loc[i,:]^T*A[i,:] using outer product */ 1415 /*-------------------------------------------------------------*/ 1416 /* put the value into Co=(p->B)^T*A (off-diagonal part, send to others) */ 1417 pnz = po->i[i+1] - po->i[i]; 1418 poJ = po->j + po->i[i]; 1419 pA = po->a + po->i[i]; 1420 for (j=0; j<pnz; j++) { 1421 row = poJ[j]; 1422 cj = coj + coi[row]; 1423 ca = coa + coi[row]; 1424 /* perform sparse axpy */ 1425 nexta = 0; 1426 valtmp = pA[j]; 1427 for (k=0; nexta<anz; k++) { 1428 if (cj[k] == adj[nexta]) { 1429 ca[k] += valtmp*ada[nexta]; 1430 nexta++; 1431 } 1432 } 1433 ierr = PetscLogFlops(2.0*anz);CHKERRQ(ierr); 1434 } 1435 1436 /* put the value into Cd (diagonal part) */ 1437 pnz = pd->i[i+1] - pd->i[i]; 1438 pdJ = pd->j + pd->i[i]; 1439 pA = pd->a + pd->i[i]; 1440 for (j=0; j<pnz; j++) { 1441 row = pdJ[j]; 1442 cj = bj + bi[row]; 1443 ca = ba + bi[row]; 1444 /* perform sparse axpy */ 1445 nexta = 0; 1446 valtmp = pA[j]; 1447 for (k=0; nexta<anz; k++) { 1448 if (cj[k] == adj[nexta]) { 1449 ca[k] += valtmp*ada[nexta]; 1450 nexta++; 1451 } 1452 } 1453 ierr = PetscLogFlops(2.0*anz);CHKERRQ(ierr); 1454 } 1455 } 1456 1457 /* 3) send and recv matrix values coa */ 1458 /*------------------------------------*/ 1459 buf_ri = merge->buf_ri; 1460 buf_rj = merge->buf_rj; 1461 len_s = merge->len_s; 1462 ierr = PetscCommGetNewTag(comm,&taga);CHKERRQ(ierr); 1463 ierr = PetscPostIrecvScalar(comm,taga,merge->nrecv,merge->id_r,merge->len_r,&abuf_r,&r_waits);CHKERRQ(ierr); 1464 1465 ierr = PetscMalloc2(merge->nsend+1,MPI_Request,&s_waits,size,MPI_Status,&status);CHKERRQ(ierr); 1466 for (proc=0,k=0; proc<size; proc++) { 1467 if (!len_s[proc]) continue; 1468 i = merge->owners_co[proc]; 1469 ierr = MPI_Isend(coa+coi[i],len_s[proc],MPIU_MATSCALAR,proc,taga,comm,s_waits+k);CHKERRQ(ierr); 1470 k++; 1471 } 1472 if (merge->nrecv) {ierr = MPI_Waitall(merge->nrecv,r_waits,status);CHKERRQ(ierr);} 1473 if (merge->nsend) {ierr = MPI_Waitall(merge->nsend,s_waits,status);CHKERRQ(ierr);} 1474 1475 ierr = PetscFree2(s_waits,status);CHKERRQ(ierr); 1476 ierr = PetscFree(r_waits);CHKERRQ(ierr); 1477 ierr = PetscFree(coa);CHKERRQ(ierr); 1478 1479 /* 4) insert local Cseq and received values into Cmpi */ 1480 /*----------------------------------------------------*/ 1481 ierr = PetscMalloc3(merge->nrecv,PetscInt**,&buf_ri_k,merge->nrecv,PetscInt*,&nextrow,merge->nrecv,PetscInt*,&nextci);CHKERRQ(ierr); 1482 for (k=0; k<merge->nrecv; k++) { 1483 buf_ri_k[k] = buf_ri[k]; /* beginning of k-th recved i-structure */ 1484 nrows = *(buf_ri_k[k]); 1485 nextrow[k] = buf_ri_k[k]+1; /* next row number of k-th recved i-structure */ 1486 nextci[k] = buf_ri_k[k] + (nrows + 1); /* poins to the next i-structure of k-th recved i-structure */ 1487 } 1488 1489 for (i=0; i<cm; i++) { 1490 row = owners[rank] + i; /* global row index of C_seq */ 1491 bj_i = bj + bi[i]; /* col indices of the i-th row of C */ 1492 ba_i = ba + bi[i]; 1493 bnz = bi[i+1] - bi[i]; 1494 /* add received vals into ba */ 1495 for (k=0; k<merge->nrecv; k++) { /* k-th received message */ 1496 /* i-th row */ 1497 if (i == *nextrow[k]) { 1498 cnz = *(nextci[k]+1) - *nextci[k]; 1499 cj = buf_rj[k] + *(nextci[k]); 1500 ca = abuf_r[k] + *(nextci[k]); 1501 nextcj = 0; 1502 for (j=0; nextcj<cnz; j++) { 1503 if (bj_i[j] == cj[nextcj]) { /* bcol == ccol */ 1504 ba_i[j] += ca[nextcj++]; 1505 } 1506 } 1507 nextrow[k]++; nextci[k]++; 1508 ierr = PetscLogFlops(2.0*cnz);CHKERRQ(ierr); 1509 } 1510 } 1511 ierr = MatSetValues(C,1,&row,bnz,bj_i,ba_i,INSERT_VALUES);CHKERRQ(ierr); 1512 } 1513 ierr = MatAssemblyBegin(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1514 ierr = MatAssemblyEnd(C,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1515 1516 ierr = PetscFree(ba);CHKERRQ(ierr); 1517 ierr = PetscFree(abuf_r[0]);CHKERRQ(ierr); 1518 ierr = PetscFree(abuf_r);CHKERRQ(ierr); 1519 ierr = PetscFree3(buf_ri_k,nextrow,nextci);CHKERRQ(ierr); 1520 PetscFunctionReturn(0); 1521 } 1522 1523 /* This routine is modified from MatPtAPSymbolic_MPIAIJ_MPIAIJ() */ 1524 #undef __FUNCT__ 1525 #define __FUNCT__ "MatTransposeMatMultSymbolic_MPIAIJ_MPIAIJ_Scalable" 1526 PetscErrorCode MatTransposeMatMultSymbolic_MPIAIJ_MPIAIJ_Scalable(Mat P,Mat A,PetscReal fill,Mat *C) 1527 { 1528 PetscErrorCode ierr; 1529 Mat Cmpi,A_loc,POt,PDt; 1530 Mat_PtAPMPI *ptap; 1531 PetscFreeSpaceList free_space=NULL,current_space=NULL; 1532 Mat_MPIAIJ *p =(Mat_MPIAIJ*)P->data,*c; 1533 PetscInt *pdti,*pdtj,*poti,*potj,*ptJ; 1534 PetscInt nnz; 1535 PetscInt *lnk,*owners_co,*coi,*coj,i,k,pnz,row; 1536 PetscInt am =A->rmap->n,pn=P->cmap->n; 1537 MPI_Comm comm; 1538 PetscMPIInt size,rank,tagi,tagj,*len_si,*len_s,*len_ri; 1539 PetscInt **buf_rj,**buf_ri,**buf_ri_k; 1540 PetscInt len,proc,*dnz,*onz,*owners; 1541 PetscInt nzi,*bi,*bj; 1542 PetscInt nrows,*buf_s,*buf_si,*buf_si_i,**nextrow,**nextci; 1543 MPI_Request *swaits,*rwaits; 1544 MPI_Status *sstatus,rstatus; 1545 Mat_Merge_SeqsToMPI *merge; 1546 PetscInt *ai,*aj,*Jptr,anz,*prmap=p->garray,pon,nspacedouble=0,j; 1547 PetscReal afill =1.0,afill_tmp; 1548 PetscInt rstart = P->cmap->rstart,rmax,aN=A->cmap->N,Crmax; 1549 PetscScalar *vals; 1550 Mat_SeqAIJ *a_loc, *pdt,*pot; 1551 1552 PetscFunctionBegin; 1553 ierr = PetscObjectGetComm((PetscObject)A,&comm);CHKERRQ(ierr); 1554 /* check if matrix local sizes are compatible */ 1555 if (A->rmap->rstart != P->rmap->rstart || A->rmap->rend != P->rmap->rend) { 1556 SETERRQ4(comm,PETSC_ERR_ARG_SIZ,"Matrix local dimensions are incompatible, A (%D, %D) != P (%D,%D)",A->rmap->rstart,A->rmap->rend,P->rmap->rstart,P->rmap->rend); 1557 } 1558 1559 ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr); 1560 ierr = MPI_Comm_rank(comm,&rank);CHKERRQ(ierr); 1561 1562 /* create struct Mat_PtAPMPI and attached it to C later */ 1563 ierr = PetscNew(Mat_PtAPMPI,&ptap);CHKERRQ(ierr); 1564 1565 /* get A_loc by taking all local rows of A */ 1566 ierr = MatMPIAIJGetLocalMat(A,MAT_INITIAL_MATRIX,&A_loc);CHKERRQ(ierr); 1567 1568 ptap->A_loc = A_loc; 1569 a_loc = (Mat_SeqAIJ*)(A_loc)->data; 1570 ai = a_loc->i; 1571 aj = a_loc->j; 1572 1573 /* determine symbolic Co=(p->B)^T*A - send to others */ 1574 /*----------------------------------------------------*/ 1575 ierr = MatTransposeSymbolic_SeqAIJ(p->A,&PDt);CHKERRQ(ierr); 1576 pdt = (Mat_SeqAIJ*)PDt->data; 1577 pdti = pdt->i; pdtj = pdt->j; 1578 1579 ierr = MatTransposeSymbolic_SeqAIJ(p->B,&POt);CHKERRQ(ierr); 1580 pot = (Mat_SeqAIJ*)POt->data; 1581 poti = pot->i; potj = pot->j; 1582 1583 /* then, compute symbolic Co = (p->B)^T*A */ 1584 pon = (p->B)->cmap->n; /* total num of rows to be sent to other processors 1585 >= (num of nonzero rows of C_seq) - pn */ 1586 ierr = PetscMalloc((pon+1)*sizeof(PetscInt),&coi);CHKERRQ(ierr); 1587 coi[0] = 0; 1588 1589 /* set initial free space to be fill*(nnz(p->B) + nnz(A)) */ 1590 nnz = fill*(poti[pon] + ai[am]); 1591 ierr = PetscFreeSpaceGet(nnz,&free_space);CHKERRQ(ierr); 1592 current_space = free_space; 1593 1594 /* create and initialize a linked list */ 1595 i = PetscMax(pdt->rmax,pot->rmax); 1596 Crmax = i*a_loc->rmax*size; /* non-scalable! */ 1597 if (!Crmax || Crmax > aN) Crmax = aN; 1598 ierr = PetscLLCondensedCreate_Scalable(Crmax,&lnk);CHKERRQ(ierr); 1599 1600 for (i=0; i<pon; i++) { 1601 pnz = poti[i+1] - poti[i]; 1602 ptJ = potj + poti[i]; 1603 for (j=0; j<pnz; j++) { 1604 row = ptJ[j]; /* row of A_loc == col of Pot */ 1605 anz = ai[row+1] - ai[row]; 1606 Jptr = aj + ai[row]; 1607 /* add non-zero cols of AP into the sorted linked list lnk */ 1608 ierr = PetscLLCondensedAddSorted_Scalable(anz,Jptr,lnk);CHKERRQ(ierr); 1609 } 1610 nnz = lnk[0]; 1611 1612 /* If free space is not available, double the total space in the list */ 1613 if (current_space->local_remaining<nnz) { 1614 ierr = PetscFreeSpaceGet(nnz+current_space->total_array_size,¤t_space);CHKERRQ(ierr); 1615 nspacedouble++; 1616 } 1617 1618 /* Copy data into free space, and zero out denserows */ 1619 ierr = PetscLLCondensedClean_Scalable(nnz,current_space->array,lnk);CHKERRQ(ierr); 1620 1621 current_space->array += nnz; 1622 current_space->local_used += nnz; 1623 current_space->local_remaining -= nnz; 1624 1625 coi[i+1] = coi[i] + nnz; 1626 } 1627 1628 ierr = PetscMalloc((coi[pon]+1)*sizeof(PetscInt),&coj);CHKERRQ(ierr); 1629 ierr = PetscFreeSpaceContiguous(&free_space,coj);CHKERRQ(ierr); 1630 1631 afill_tmp = (PetscReal)coi[pon]/(poti[pon] + ai[am]+1); 1632 if (afill_tmp > afill) afill = afill_tmp; 1633 1634 /* send j-array (coj) of Co to other processors */ 1635 /*----------------------------------------------*/ 1636 /* determine row ownership */ 1637 ierr = PetscNew(Mat_Merge_SeqsToMPI,&merge);CHKERRQ(ierr); 1638 ierr = PetscLayoutCreate(comm,&merge->rowmap);CHKERRQ(ierr); 1639 1640 merge->rowmap->n = pn; 1641 merge->rowmap->bs = 1; 1642 1643 ierr = PetscLayoutSetUp(merge->rowmap);CHKERRQ(ierr); 1644 owners = merge->rowmap->range; 1645 1646 /* determine the number of messages to send, their lengths */ 1647 ierr = PetscMalloc(size*sizeof(PetscMPIInt),&len_si);CHKERRQ(ierr); 1648 ierr = PetscMemzero(len_si,size*sizeof(PetscMPIInt));CHKERRQ(ierr); 1649 ierr = PetscMalloc(size*sizeof(PetscMPIInt),&merge->len_s);CHKERRQ(ierr); 1650 1651 len_s = merge->len_s; 1652 merge->nsend = 0; 1653 1654 ierr = PetscMalloc((size+2)*sizeof(PetscInt),&owners_co);CHKERRQ(ierr); 1655 ierr = PetscMemzero(len_s,size*sizeof(PetscMPIInt));CHKERRQ(ierr); 1656 1657 proc = 0; 1658 for (i=0; i<pon; i++) { 1659 while (prmap[i] >= owners[proc+1]) proc++; 1660 len_si[proc]++; /* num of rows in Co to be sent to [proc] */ 1661 len_s[proc] += coi[i+1] - coi[i]; 1662 } 1663 1664 len = 0; /* max length of buf_si[] */ 1665 owners_co[0] = 0; 1666 for (proc=0; proc<size; proc++) { 1667 owners_co[proc+1] = owners_co[proc] + len_si[proc]; 1668 if (len_si[proc]) { 1669 merge->nsend++; 1670 len_si[proc] = 2*(len_si[proc] + 1); 1671 len += len_si[proc]; 1672 } 1673 } 1674 1675 /* determine the number and length of messages to receive for coi and coj */ 1676 ierr = PetscGatherNumberOfMessages(comm,NULL,len_s,&merge->nrecv);CHKERRQ(ierr); 1677 ierr = PetscGatherMessageLengths2(comm,merge->nsend,merge->nrecv,len_s,len_si,&merge->id_r,&merge->len_r,&len_ri);CHKERRQ(ierr); 1678 1679 /* post the Irecv and Isend of coj */ 1680 ierr = PetscCommGetNewTag(comm,&tagj);CHKERRQ(ierr); 1681 ierr = PetscPostIrecvInt(comm,tagj,merge->nrecv,merge->id_r,merge->len_r,&buf_rj,&rwaits);CHKERRQ(ierr); 1682 ierr = PetscMalloc((merge->nsend+1)*sizeof(MPI_Request),&swaits);CHKERRQ(ierr); 1683 for (proc=0, k=0; proc<size; proc++) { 1684 if (!len_s[proc]) continue; 1685 i = owners_co[proc]; 1686 ierr = MPI_Isend(coj+coi[i],len_s[proc],MPIU_INT,proc,tagj,comm,swaits+k);CHKERRQ(ierr); 1687 k++; 1688 } 1689 1690 /* receives and sends of coj are complete */ 1691 ierr = PetscMalloc(size*sizeof(MPI_Status),&sstatus);CHKERRQ(ierr); 1692 for (i=0; i<merge->nrecv; i++) { 1693 PetscMPIInt icompleted; 1694 ierr = MPI_Waitany(merge->nrecv,rwaits,&icompleted,&rstatus);CHKERRQ(ierr); 1695 } 1696 ierr = PetscFree(rwaits);CHKERRQ(ierr); 1697 if (merge->nsend) {ierr = MPI_Waitall(merge->nsend,swaits,sstatus);CHKERRQ(ierr);} 1698 1699 /* send and recv coi */ 1700 /*-------------------*/ 1701 ierr = PetscCommGetNewTag(comm,&tagi);CHKERRQ(ierr); 1702 ierr = PetscPostIrecvInt(comm,tagi,merge->nrecv,merge->id_r,len_ri,&buf_ri,&rwaits);CHKERRQ(ierr); 1703 ierr = PetscMalloc((len+1)*sizeof(PetscInt),&buf_s);CHKERRQ(ierr); 1704 buf_si = buf_s; /* points to the beginning of k-th msg to be sent */ 1705 for (proc=0,k=0; proc<size; proc++) { 1706 if (!len_s[proc]) continue; 1707 /* form outgoing message for i-structure: 1708 buf_si[0]: nrows to be sent 1709 [1:nrows]: row index (global) 1710 [nrows+1:2*nrows+1]: i-structure index 1711 */ 1712 /*-------------------------------------------*/ 1713 nrows = len_si[proc]/2 - 1; 1714 buf_si_i = buf_si + nrows+1; 1715 buf_si[0] = nrows; 1716 buf_si_i[0] = 0; 1717 nrows = 0; 1718 for (i=owners_co[proc]; i<owners_co[proc+1]; i++) { 1719 nzi = coi[i+1] - coi[i]; 1720 buf_si_i[nrows+1] = buf_si_i[nrows] + nzi; /* i-structure */ 1721 buf_si[nrows+1] = prmap[i] -owners[proc]; /* local row index */ 1722 nrows++; 1723 } 1724 ierr = MPI_Isend(buf_si,len_si[proc],MPIU_INT,proc,tagi,comm,swaits+k);CHKERRQ(ierr); 1725 k++; 1726 buf_si += len_si[proc]; 1727 } 1728 i = merge->nrecv; 1729 while (i--) { 1730 PetscMPIInt icompleted; 1731 ierr = MPI_Waitany(merge->nrecv,rwaits,&icompleted,&rstatus);CHKERRQ(ierr); 1732 } 1733 ierr = PetscFree(rwaits);CHKERRQ(ierr); 1734 if (merge->nsend) {ierr = MPI_Waitall(merge->nsend,swaits,sstatus);CHKERRQ(ierr);} 1735 ierr = PetscFree(len_si);CHKERRQ(ierr); 1736 ierr = PetscFree(len_ri);CHKERRQ(ierr); 1737 ierr = PetscFree(swaits);CHKERRQ(ierr); 1738 ierr = PetscFree(sstatus);CHKERRQ(ierr); 1739 ierr = PetscFree(buf_s);CHKERRQ(ierr); 1740 1741 /* compute the local portion of C (mpi mat) */ 1742 /*------------------------------------------*/ 1743 /* allocate bi array and free space for accumulating nonzero column info */ 1744 ierr = PetscMalloc((pn+1)*sizeof(PetscInt),&bi);CHKERRQ(ierr); 1745 bi[0] = 0; 1746 1747 /* set initial free space to be fill*(nnz(P) + nnz(AP)) */ 1748 nnz = fill*(pdti[pn] + poti[pon] + ai[am]); 1749 ierr = PetscFreeSpaceGet(nnz,&free_space);CHKERRQ(ierr); 1750 current_space = free_space; 1751 1752 ierr = PetscMalloc3(merge->nrecv,PetscInt**,&buf_ri_k,merge->nrecv,PetscInt*,&nextrow,merge->nrecv,PetscInt*,&nextci);CHKERRQ(ierr); 1753 for (k=0; k<merge->nrecv; k++) { 1754 buf_ri_k[k] = buf_ri[k]; /* beginning of k-th recved i-structure */ 1755 nrows = *buf_ri_k[k]; 1756 nextrow[k] = buf_ri_k[k] + 1; /* next row number of k-th recved i-structure */ 1757 nextci[k] = buf_ri_k[k] + (nrows + 1); /* points to the next i-structure of k-th recieved i-structure */ 1758 } 1759 1760 ierr = MatPreallocateInitialize(comm,pn,A->cmap->n,dnz,onz);CHKERRQ(ierr); 1761 rmax = 0; 1762 for (i=0; i<pn; i++) { 1763 /* add pdt[i,:]*AP into lnk */ 1764 pnz = pdti[i+1] - pdti[i]; 1765 ptJ = pdtj + pdti[i]; 1766 for (j=0; j<pnz; j++) { 1767 row = ptJ[j]; /* row of AP == col of Pt */ 1768 anz = ai[row+1] - ai[row]; 1769 Jptr = aj + ai[row]; 1770 /* add non-zero cols of AP into the sorted linked list lnk */ 1771 ierr = PetscLLCondensedAddSorted_Scalable(anz,Jptr,lnk);CHKERRQ(ierr); 1772 } 1773 1774 /* add received col data into lnk */ 1775 for (k=0; k<merge->nrecv; k++) { /* k-th received message */ 1776 if (i == *nextrow[k]) { /* i-th row */ 1777 nzi = *(nextci[k]+1) - *nextci[k]; 1778 Jptr = buf_rj[k] + *nextci[k]; 1779 ierr = PetscLLCondensedAddSorted_Scalable(nzi,Jptr,lnk);CHKERRQ(ierr); 1780 nextrow[k]++; nextci[k]++; 1781 } 1782 } 1783 nnz = lnk[0]; 1784 1785 /* if free space is not available, make more free space */ 1786 if (current_space->local_remaining<nnz) { 1787 ierr = PetscFreeSpaceGet(nnz+current_space->total_array_size,¤t_space);CHKERRQ(ierr); 1788 nspacedouble++; 1789 } 1790 /* copy data into free space, then initialize lnk */ 1791 ierr = PetscLLCondensedClean_Scalable(nnz,current_space->array,lnk);CHKERRQ(ierr); 1792 ierr = MatPreallocateSet(i+owners[rank],nnz,current_space->array,dnz,onz);CHKERRQ(ierr); 1793 1794 current_space->array += nnz; 1795 current_space->local_used += nnz; 1796 current_space->local_remaining -= nnz; 1797 1798 bi[i+1] = bi[i] + nnz; 1799 if (nnz > rmax) rmax = nnz; 1800 } 1801 ierr = PetscFree3(buf_ri_k,nextrow,nextci);CHKERRQ(ierr); 1802 1803 ierr = PetscMalloc((bi[pn]+1)*sizeof(PetscInt),&bj);CHKERRQ(ierr); 1804 ierr = PetscFreeSpaceContiguous(&free_space,bj);CHKERRQ(ierr); 1805 afill_tmp = (PetscReal)bi[pn]/(pdti[pn] + poti[pon] + ai[am]+1); 1806 if (afill_tmp > afill) afill = afill_tmp; 1807 ierr = PetscLLCondensedDestroy_Scalable(lnk);CHKERRQ(ierr); 1808 ierr = MatDestroy(&POt);CHKERRQ(ierr); 1809 ierr = MatDestroy(&PDt);CHKERRQ(ierr); 1810 1811 /* create symbolic parallel matrix Cmpi - why cannot be assembled in Numeric part */ 1812 /*----------------------------------------------------------------------------------*/ 1813 ierr = PetscMalloc((rmax+1)*sizeof(PetscScalar),&vals);CHKERRQ(ierr); 1814 ierr = PetscMemzero(vals,rmax*sizeof(PetscScalar));CHKERRQ(ierr); 1815 1816 ierr = MatCreate(comm,&Cmpi);CHKERRQ(ierr); 1817 ierr = MatSetSizes(Cmpi,pn,A->cmap->n,PETSC_DETERMINE,PETSC_DETERMINE);CHKERRQ(ierr); 1818 ierr = MatSetBlockSizes(Cmpi,P->cmap->bs,A->cmap->bs);CHKERRQ(ierr); 1819 ierr = MatSetType(Cmpi,MATMPIAIJ);CHKERRQ(ierr); 1820 ierr = MatMPIAIJSetPreallocation(Cmpi,0,dnz,0,onz);CHKERRQ(ierr); 1821 ierr = MatPreallocateFinalize(dnz,onz);CHKERRQ(ierr); 1822 ierr = MatSetBlockSize(Cmpi,1);CHKERRQ(ierr); 1823 for (i=0; i<pn; i++) { 1824 row = i + rstart; 1825 nnz = bi[i+1] - bi[i]; 1826 Jptr = bj + bi[i]; 1827 ierr = MatSetValues(Cmpi,1,&row,nnz,Jptr,vals,INSERT_VALUES);CHKERRQ(ierr); 1828 } 1829 ierr = MatAssemblyBegin(Cmpi,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1830 ierr = MatAssemblyEnd(Cmpi,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 1831 ierr = PetscFree(vals);CHKERRQ(ierr); 1832 1833 merge->bi = bi; 1834 merge->bj = bj; 1835 merge->coi = coi; 1836 merge->coj = coj; 1837 merge->buf_ri = buf_ri; 1838 merge->buf_rj = buf_rj; 1839 merge->owners_co = owners_co; 1840 merge->destroy = Cmpi->ops->destroy; 1841 merge->duplicate = Cmpi->ops->duplicate; 1842 1843 Cmpi->ops->mattransposemultnumeric = MatTransposeMatMultNumeric_MPIAIJ_MPIAIJ_Scalable; 1844 Cmpi->ops->destroy = MatDestroy_MPIAIJ_PtAP; 1845 1846 /* attach the supporting struct to Cmpi for reuse */ 1847 c = (Mat_MPIAIJ*)Cmpi->data; 1848 1849 c->ptap = ptap; 1850 ptap->api = NULL; 1851 ptap->apj = NULL; 1852 ptap->merge = merge; 1853 ptap->rmax = rmax; 1854 ptap->apa = NULL; 1855 1856 *C = Cmpi; 1857 #if defined(PETSC_USE_INFO) 1858 if (bi[pn] != 0) { 1859 ierr = PetscInfo3(Cmpi,"Reallocs %D; Fill ratio: given %G needed %G.\n",nspacedouble,fill,afill);CHKERRQ(ierr); 1860 ierr = PetscInfo1(Cmpi,"Use MatTransposeMatMult(A,B,MatReuse,%G,&C) for best performance.\n",afill);CHKERRQ(ierr); 1861 } else { 1862 ierr = PetscInfo(Cmpi,"Empty matrix product\n");CHKERRQ(ierr); 1863 } 1864 #endif 1865 PetscFunctionReturn(0); 1866 } 1867