1 #define PETSCMAT_DLL 2 3 #include "src/mat/impls/baij/mpi/mpibaij.h" /*I "petscmat.h" I*/ 4 #include "mpisbaij.h" 5 #include "src/mat/impls/sbaij/seq/sbaij.h" 6 7 EXTERN PetscErrorCode MatSetUpMultiply_MPISBAIJ(Mat); 8 EXTERN PetscErrorCode MatSetUpMultiply_MPISBAIJ_2comm(Mat); 9 EXTERN PetscErrorCode DisAssemble_MPISBAIJ(Mat); 10 EXTERN PetscErrorCode MatIncreaseOverlap_MPISBAIJ(Mat,PetscInt,IS[],PetscInt); 11 EXTERN PetscErrorCode MatGetValues_SeqSBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],PetscScalar []); 12 EXTERN PetscErrorCode MatGetValues_SeqBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],PetscScalar []); 13 EXTERN PetscErrorCode MatSetValues_SeqSBAIJ(Mat,PetscInt,const PetscInt [],PetscInt,const PetscInt [],const PetscScalar [],InsertMode); 14 EXTERN PetscErrorCode MatSetValuesBlocked_SeqSBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const PetscScalar[],InsertMode); 15 EXTERN PetscErrorCode MatSetValuesBlocked_SeqBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const PetscScalar[],InsertMode); 16 EXTERN PetscErrorCode MatGetRow_SeqSBAIJ(Mat,PetscInt,PetscInt*,PetscInt**,PetscScalar**); 17 EXTERN PetscErrorCode MatRestoreRow_SeqSBAIJ(Mat,PetscInt,PetscInt*,PetscInt**,PetscScalar**); 18 EXTERN PetscErrorCode MatZeroRows_SeqSBAIJ(Mat,IS,PetscScalar*); 19 EXTERN PetscErrorCode MatZeroRows_SeqBAIJ(Mat,IS,PetscScalar *); 20 EXTERN PetscErrorCode MatGetRowMaxAbs_MPISBAIJ(Mat,Vec,PetscInt[]); 21 EXTERN PetscErrorCode MatRelax_MPISBAIJ(Mat,Vec,PetscReal,MatSORType,PetscReal,PetscInt,PetscInt,Vec); 22 23 EXTERN_C_BEGIN 24 #undef __FUNCT__ 25 #define __FUNCT__ "MatStoreValues_MPISBAIJ" 26 PetscErrorCode PETSCMAT_DLLEXPORT MatStoreValues_MPISBAIJ(Mat mat) 27 { 28 Mat_MPISBAIJ *aij = (Mat_MPISBAIJ *)mat->data; 29 PetscErrorCode ierr; 30 31 PetscFunctionBegin; 32 ierr = MatStoreValues(aij->A);CHKERRQ(ierr); 33 ierr = MatStoreValues(aij->B);CHKERRQ(ierr); 34 PetscFunctionReturn(0); 35 } 36 EXTERN_C_END 37 38 EXTERN_C_BEGIN 39 #undef __FUNCT__ 40 #define __FUNCT__ "MatRetrieveValues_MPISBAIJ" 41 PetscErrorCode PETSCMAT_DLLEXPORT MatRetrieveValues_MPISBAIJ(Mat mat) 42 { 43 Mat_MPISBAIJ *aij = (Mat_MPISBAIJ *)mat->data; 44 PetscErrorCode ierr; 45 46 PetscFunctionBegin; 47 ierr = MatRetrieveValues(aij->A);CHKERRQ(ierr); 48 ierr = MatRetrieveValues(aij->B);CHKERRQ(ierr); 49 PetscFunctionReturn(0); 50 } 51 EXTERN_C_END 52 53 54 #define CHUNKSIZE 10 55 56 #define MatSetValues_SeqSBAIJ_A_Private(row,col,value,addv) \ 57 { \ 58 \ 59 brow = row/bs; \ 60 rp = aj + ai[brow]; ap = aa + bs2*ai[brow]; \ 61 rmax = aimax[brow]; nrow = ailen[brow]; \ 62 bcol = col/bs; \ 63 ridx = row % bs; cidx = col % bs; \ 64 low = 0; high = nrow; \ 65 while (high-low > 3) { \ 66 t = (low+high)/2; \ 67 if (rp[t] > bcol) high = t; \ 68 else low = t; \ 69 } \ 70 for (_i=low; _i<high; _i++) { \ 71 if (rp[_i] > bcol) break; \ 72 if (rp[_i] == bcol) { \ 73 bap = ap + bs2*_i + bs*cidx + ridx; \ 74 if (addv == ADD_VALUES) *bap += value; \ 75 else *bap = value; \ 76 goto a_noinsert; \ 77 } \ 78 } \ 79 if (a->nonew == 1) goto a_noinsert; \ 80 if (a->nonew == -1) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) into matrix", row, col); \ 81 MatSeqXAIJReallocateAIJ(A,a->mbs,bs2,nrow,brow,bcol,rmax,aa,ai,aj,rp,ap,aimax,a->nonew,MatScalar); \ 82 N = nrow++ - 1; \ 83 /* shift up all the later entries in this row */ \ 84 for (ii=N; ii>=_i; ii--) { \ 85 rp[ii+1] = rp[ii]; \ 86 ierr = PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar));CHKERRQ(ierr); \ 87 } \ 88 if (N>=_i) { ierr = PetscMemzero(ap+bs2*_i,bs2*sizeof(MatScalar));CHKERRQ(ierr); } \ 89 rp[_i] = bcol; \ 90 ap[bs2*_i + bs*cidx + ridx] = value; \ 91 a_noinsert:; \ 92 ailen[brow] = nrow; \ 93 } 94 95 #define MatSetValues_SeqSBAIJ_B_Private(row,col,value,addv) \ 96 { \ 97 brow = row/bs; \ 98 rp = bj + bi[brow]; ap = ba + bs2*bi[brow]; \ 99 rmax = bimax[brow]; nrow = bilen[brow]; \ 100 bcol = col/bs; \ 101 ridx = row % bs; cidx = col % bs; \ 102 low = 0; high = nrow; \ 103 while (high-low > 3) { \ 104 t = (low+high)/2; \ 105 if (rp[t] > bcol) high = t; \ 106 else low = t; \ 107 } \ 108 for (_i=low; _i<high; _i++) { \ 109 if (rp[_i] > bcol) break; \ 110 if (rp[_i] == bcol) { \ 111 bap = ap + bs2*_i + bs*cidx + ridx; \ 112 if (addv == ADD_VALUES) *bap += value; \ 113 else *bap = value; \ 114 goto b_noinsert; \ 115 } \ 116 } \ 117 if (b->nonew == 1) goto b_noinsert; \ 118 if (b->nonew == -1) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Inserting a new nonzero (%D, %D) into matrix", row, col); \ 119 MatSeqXAIJReallocateAIJ(B,b->mbs,bs2,nrow,brow,bcol,rmax,ba,bi,bj,rp,ap,bimax,b->nonew,MatScalar); \ 120 N = nrow++ - 1; \ 121 /* shift up all the later entries in this row */ \ 122 for (ii=N; ii>=_i; ii--) { \ 123 rp[ii+1] = rp[ii]; \ 124 ierr = PetscMemcpy(ap+bs2*(ii+1),ap+bs2*(ii),bs2*sizeof(MatScalar));CHKERRQ(ierr); \ 125 } \ 126 if (N>=_i) { ierr = PetscMemzero(ap+bs2*_i,bs2*sizeof(MatScalar));CHKERRQ(ierr);} \ 127 rp[_i] = bcol; \ 128 ap[bs2*_i + bs*cidx + ridx] = value; \ 129 b_noinsert:; \ 130 bilen[brow] = nrow; \ 131 } 132 133 /* Only add/insert a(i,j) with i<=j (blocks). 134 Any a(i,j) with i>j input by user is ingored. 135 */ 136 #undef __FUNCT__ 137 #define __FUNCT__ "MatSetValues_MPISBAIJ" 138 PetscErrorCode MatSetValues_MPISBAIJ(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const PetscScalar v[],InsertMode addv) 139 { 140 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 141 MatScalar value; 142 PetscTruth roworiented = baij->roworiented; 143 PetscErrorCode ierr; 144 PetscInt i,j,row,col; 145 PetscInt rstart_orig=mat->rmap.rstart; 146 PetscInt rend_orig=mat->rmap.rend,cstart_orig=mat->cmap.rstart; 147 PetscInt cend_orig=mat->cmap.rend,bs=mat->rmap.bs; 148 149 /* Some Variables required in the macro */ 150 Mat A = baij->A; 151 Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)(A)->data; 152 PetscInt *aimax=a->imax,*ai=a->i,*ailen=a->ilen,*aj=a->j; 153 MatScalar *aa=a->a; 154 155 Mat B = baij->B; 156 Mat_SeqBAIJ *b = (Mat_SeqBAIJ*)(B)->data; 157 PetscInt *bimax=b->imax,*bi=b->i,*bilen=b->ilen,*bj=b->j; 158 MatScalar *ba=b->a; 159 160 PetscInt *rp,ii,nrow,_i,rmax,N,brow,bcol; 161 PetscInt low,high,t,ridx,cidx,bs2=a->bs2; 162 MatScalar *ap,*bap; 163 164 /* for stash */ 165 PetscInt n_loc, *in_loc = PETSC_NULL; 166 MatScalar *v_loc = PETSC_NULL; 167 168 PetscFunctionBegin; 169 170 if (!baij->donotstash){ 171 if (n > baij->n_loc) { 172 ierr = PetscFree(baij->in_loc);CHKERRQ(ierr); 173 ierr = PetscFree(baij->v_loc);CHKERRQ(ierr); 174 ierr = PetscMalloc(n*sizeof(PetscInt),&baij->in_loc);CHKERRQ(ierr); 175 ierr = PetscMalloc(n*sizeof(MatScalar),&baij->v_loc);CHKERRQ(ierr); 176 baij->n_loc = n; 177 } 178 in_loc = baij->in_loc; 179 v_loc = baij->v_loc; 180 } 181 182 for (i=0; i<m; i++) { 183 if (im[i] < 0) continue; 184 #if defined(PETSC_USE_DEBUG) 185 if (im[i] >= mat->rmap.N) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",im[i],mat->rmap.N-1); 186 #endif 187 if (im[i] >= rstart_orig && im[i] < rend_orig) { /* this processor entry */ 188 row = im[i] - rstart_orig; /* local row index */ 189 for (j=0; j<n; j++) { 190 if (im[i]/bs > in[j]/bs){ 191 if (a->ignore_ltriangular){ 192 continue; /* ignore lower triangular blocks */ 193 } else { 194 SETERRQ(PETSC_ERR_USER,"Lower triangular value cannot be set for sbaij format. Ignoring these values, run with -mat_ignore_lower_triangular or call MatSetOption(mat,MAT_IGNORE_LOWER_TRIANGULAR,PETSC_TRUE)"); 195 } 196 } 197 if (in[j] >= cstart_orig && in[j] < cend_orig){ /* diag entry (A) */ 198 col = in[j] - cstart_orig; /* local col index */ 199 brow = row/bs; bcol = col/bs; 200 if (brow > bcol) continue; /* ignore lower triangular blocks of A */ 201 if (roworiented) value = v[i*n+j]; else value = v[i+j*m]; 202 MatSetValues_SeqSBAIJ_A_Private(row,col,value,addv); 203 /* ierr = MatSetValues_SeqBAIJ(baij->A,1,&row,1,&col,&value,addv);CHKERRQ(ierr); */ 204 } else if (in[j] < 0) continue; 205 #if defined(PETSC_USE_DEBUG) 206 else if (in[j] >= mat->cmap.N) {SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",in[j],mat->cmap.N-1);} 207 #endif 208 else { /* off-diag entry (B) */ 209 if (mat->was_assembled) { 210 if (!baij->colmap) { 211 ierr = CreateColmap_MPIBAIJ_Private(mat);CHKERRQ(ierr); 212 } 213 #if defined (PETSC_USE_CTABLE) 214 ierr = PetscTableFind(baij->colmap,in[j]/bs + 1,&col);CHKERRQ(ierr); 215 col = col - 1; 216 #else 217 col = baij->colmap[in[j]/bs] - 1; 218 #endif 219 if (col < 0 && !((Mat_SeqSBAIJ*)(baij->A->data))->nonew) { 220 ierr = DisAssemble_MPISBAIJ(mat);CHKERRQ(ierr); 221 col = in[j]; 222 /* Reinitialize the variables required by MatSetValues_SeqBAIJ_B_Private() */ 223 B = baij->B; 224 b = (Mat_SeqBAIJ*)(B)->data; 225 bimax=b->imax;bi=b->i;bilen=b->ilen;bj=b->j; 226 ba=b->a; 227 } else col += in[j]%bs; 228 } else col = in[j]; 229 if (roworiented) value = v[i*n+j]; else value = v[i+j*m]; 230 MatSetValues_SeqSBAIJ_B_Private(row,col,value,addv); 231 /* ierr = MatSetValues_SeqBAIJ(baij->B,1,&row,1,&col,&value,addv);CHKERRQ(ierr); */ 232 } 233 } 234 } else { /* off processor entry */ 235 if (!baij->donotstash) { 236 n_loc = 0; 237 for (j=0; j<n; j++){ 238 if (im[i]/bs > in[j]/bs) continue; /* ignore lower triangular blocks */ 239 in_loc[n_loc] = in[j]; 240 if (roworiented) { 241 v_loc[n_loc] = v[i*n+j]; 242 } else { 243 v_loc[n_loc] = v[j*m+i]; 244 } 245 n_loc++; 246 } 247 ierr = MatStashValuesRow_Private(&mat->stash,im[i],n_loc,in_loc,v_loc);CHKERRQ(ierr); 248 } 249 } 250 } 251 PetscFunctionReturn(0); 252 } 253 254 #undef __FUNCT__ 255 #define __FUNCT__ "MatSetValuesBlocked_MPISBAIJ" 256 PetscErrorCode MatSetValuesBlocked_MPISBAIJ(Mat mat,PetscInt m,const PetscInt im[],PetscInt n,const PetscInt in[],const MatScalar v[],InsertMode addv) 257 { 258 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 259 const MatScalar *value; 260 MatScalar *barray=baij->barray; 261 PetscTruth roworiented = baij->roworiented; 262 PetscErrorCode ierr; 263 PetscInt i,j,ii,jj,row,col,rstart=baij->rstartbs; 264 PetscInt rend=baij->rendbs,cstart=baij->rstartbs,stepval; 265 PetscInt cend=baij->rendbs,bs=mat->rmap.bs,bs2=baij->bs2; 266 267 PetscFunctionBegin; 268 if(!barray) { 269 ierr = PetscMalloc(bs2*sizeof(MatScalar),&barray);CHKERRQ(ierr); 270 baij->barray = barray; 271 } 272 273 if (roworiented) { 274 stepval = (n-1)*bs; 275 } else { 276 stepval = (m-1)*bs; 277 } 278 for (i=0; i<m; i++) { 279 if (im[i] < 0) continue; 280 #if defined(PETSC_USE_DEBUG) 281 if (im[i] >= baij->Mbs) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large, row %D max %D",im[i],baij->Mbs-1); 282 #endif 283 if (im[i] >= rstart && im[i] < rend) { 284 row = im[i] - rstart; 285 for (j=0; j<n; j++) { 286 /* If NumCol = 1 then a copy is not required */ 287 if ((roworiented) && (n == 1)) { 288 barray = (MatScalar*) v + i*bs2; 289 } else if((!roworiented) && (m == 1)) { 290 barray = (MatScalar*) v + j*bs2; 291 } else { /* Here a copy is required */ 292 if (roworiented) { 293 value = v + i*(stepval+bs)*bs + j*bs; 294 } else { 295 value = v + j*(stepval+bs)*bs + i*bs; 296 } 297 for (ii=0; ii<bs; ii++,value+=stepval) { 298 for (jj=0; jj<bs; jj++) { 299 *barray++ = *value++; 300 } 301 } 302 barray -=bs2; 303 } 304 305 if (in[j] >= cstart && in[j] < cend){ 306 col = in[j] - cstart; 307 ierr = MatSetValuesBlocked_SeqSBAIJ(baij->A,1,&row,1,&col,barray,addv);CHKERRQ(ierr); 308 } 309 else if (in[j] < 0) continue; 310 #if defined(PETSC_USE_DEBUG) 311 else if (in[j] >= baij->Nbs) {SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column too large, col %D max %D",in[j],baij->Nbs-1);} 312 #endif 313 else { 314 if (mat->was_assembled) { 315 if (!baij->colmap) { 316 ierr = CreateColmap_MPIBAIJ_Private(mat);CHKERRQ(ierr); 317 } 318 319 #if defined(PETSC_USE_DEBUG) 320 #if defined (PETSC_USE_CTABLE) 321 { PetscInt data; 322 ierr = PetscTableFind(baij->colmap,in[j]+1,&data);CHKERRQ(ierr); 323 if ((data - 1) % bs) SETERRQ(PETSC_ERR_PLIB,"Incorrect colmap"); 324 } 325 #else 326 if ((baij->colmap[in[j]] - 1) % bs) SETERRQ(PETSC_ERR_PLIB,"Incorrect colmap"); 327 #endif 328 #endif 329 #if defined (PETSC_USE_CTABLE) 330 ierr = PetscTableFind(baij->colmap,in[j]+1,&col);CHKERRQ(ierr); 331 col = (col - 1)/bs; 332 #else 333 col = (baij->colmap[in[j]] - 1)/bs; 334 #endif 335 if (col < 0 && !((Mat_SeqBAIJ*)(baij->A->data))->nonew) { 336 ierr = DisAssemble_MPISBAIJ(mat);CHKERRQ(ierr); 337 col = in[j]; 338 } 339 } 340 else col = in[j]; 341 ierr = MatSetValuesBlocked_SeqBAIJ(baij->B,1,&row,1,&col,barray,addv);CHKERRQ(ierr); 342 } 343 } 344 } else { 345 if (!baij->donotstash) { 346 if (roworiented) { 347 ierr = MatStashValuesRowBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);CHKERRQ(ierr); 348 } else { 349 ierr = MatStashValuesColBlocked_Private(&mat->bstash,im[i],n,in,v,m,n,i);CHKERRQ(ierr); 350 } 351 } 352 } 353 } 354 PetscFunctionReturn(0); 355 } 356 357 #undef __FUNCT__ 358 #define __FUNCT__ "MatGetValues_MPISBAIJ" 359 PetscErrorCode MatGetValues_MPISBAIJ(Mat mat,PetscInt m,const PetscInt idxm[],PetscInt n,const PetscInt idxn[],PetscScalar v[]) 360 { 361 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 362 PetscErrorCode ierr; 363 PetscInt bs=mat->rmap.bs,i,j,bsrstart = mat->rmap.rstart,bsrend = mat->rmap.rend; 364 PetscInt bscstart = mat->cmap.rstart,bscend = mat->cmap.rend,row,col,data; 365 366 PetscFunctionBegin; 367 for (i=0; i<m; i++) { 368 if (idxm[i] < 0) continue; /* SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"Negative row: %D",idxm[i]); */ 369 if (idxm[i] >= mat->rmap.N) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Row too large: row %D max %D",idxm[i],mat->rmap.N-1); 370 if (idxm[i] >= bsrstart && idxm[i] < bsrend) { 371 row = idxm[i] - bsrstart; 372 for (j=0; j<n; j++) { 373 if (idxn[j] < 0) continue; /* SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"Negative column %D",idxn[j]); */ 374 if (idxn[j] >= mat->cmap.N) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"Column too large: col %D max %D",idxn[j],mat->cmap.N-1); 375 if (idxn[j] >= bscstart && idxn[j] < bscend){ 376 col = idxn[j] - bscstart; 377 ierr = MatGetValues_SeqSBAIJ(baij->A,1,&row,1,&col,v+i*n+j);CHKERRQ(ierr); 378 } else { 379 if (!baij->colmap) { 380 ierr = CreateColmap_MPIBAIJ_Private(mat);CHKERRQ(ierr); 381 } 382 #if defined (PETSC_USE_CTABLE) 383 ierr = PetscTableFind(baij->colmap,idxn[j]/bs+1,&data);CHKERRQ(ierr); 384 data --; 385 #else 386 data = baij->colmap[idxn[j]/bs]-1; 387 #endif 388 if((data < 0) || (baij->garray[data/bs] != idxn[j]/bs)) *(v+i*n+j) = 0.0; 389 else { 390 col = data + idxn[j]%bs; 391 ierr = MatGetValues_SeqBAIJ(baij->B,1,&row,1,&col,v+i*n+j);CHKERRQ(ierr); 392 } 393 } 394 } 395 } else { 396 SETERRQ(PETSC_ERR_SUP,"Only local values currently supported"); 397 } 398 } 399 PetscFunctionReturn(0); 400 } 401 402 #undef __FUNCT__ 403 #define __FUNCT__ "MatNorm_MPISBAIJ" 404 PetscErrorCode MatNorm_MPISBAIJ(Mat mat,NormType type,PetscReal *norm) 405 { 406 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 407 PetscErrorCode ierr; 408 PetscReal sum[2],*lnorm2; 409 410 PetscFunctionBegin; 411 if (baij->size == 1) { 412 ierr = MatNorm(baij->A,type,norm);CHKERRQ(ierr); 413 } else { 414 if (type == NORM_FROBENIUS) { 415 ierr = PetscMalloc(2*sizeof(PetscReal),&lnorm2);CHKERRQ(ierr); 416 ierr = MatNorm(baij->A,type,lnorm2);CHKERRQ(ierr); 417 *lnorm2 = (*lnorm2)*(*lnorm2); lnorm2++; /* squar power of norm(A) */ 418 ierr = MatNorm(baij->B,type,lnorm2);CHKERRQ(ierr); 419 *lnorm2 = (*lnorm2)*(*lnorm2); lnorm2--; /* squar power of norm(B) */ 420 ierr = MPI_Allreduce(lnorm2,&sum,2,MPIU_REAL,MPI_SUM,((PetscObject)mat)->comm);CHKERRQ(ierr); 421 *norm = sqrt(sum[0] + 2*sum[1]); 422 ierr = PetscFree(lnorm2);CHKERRQ(ierr); 423 } else if (type == NORM_INFINITY || type == NORM_1) { /* max row/column sum */ 424 Mat_SeqSBAIJ *amat=(Mat_SeqSBAIJ*)baij->A->data; 425 Mat_SeqBAIJ *bmat=(Mat_SeqBAIJ*)baij->B->data; 426 PetscReal *rsum,*rsum2,vabs; 427 PetscInt *jj,*garray=baij->garray,rstart=baij->rstartbs,nz; 428 PetscInt brow,bcol,col,bs=baij->A->rmap.bs,row,grow,gcol,mbs=amat->mbs; 429 MatScalar *v; 430 431 ierr = PetscMalloc((2*mat->cmap.N+1)*sizeof(PetscReal),&rsum);CHKERRQ(ierr); 432 rsum2 = rsum + mat->cmap.N; 433 ierr = PetscMemzero(rsum,mat->cmap.N*sizeof(PetscReal));CHKERRQ(ierr); 434 /* Amat */ 435 v = amat->a; jj = amat->j; 436 for (brow=0; brow<mbs; brow++) { 437 grow = bs*(rstart + brow); 438 nz = amat->i[brow+1] - amat->i[brow]; 439 for (bcol=0; bcol<nz; bcol++){ 440 gcol = bs*(rstart + *jj); jj++; 441 for (col=0; col<bs; col++){ 442 for (row=0; row<bs; row++){ 443 vabs = PetscAbsScalar(*v); v++; 444 rsum[gcol+col] += vabs; 445 /* non-diagonal block */ 446 if (bcol > 0 && vabs > 0.0) rsum[grow+row] += vabs; 447 } 448 } 449 } 450 } 451 /* Bmat */ 452 v = bmat->a; jj = bmat->j; 453 for (brow=0; brow<mbs; brow++) { 454 grow = bs*(rstart + brow); 455 nz = bmat->i[brow+1] - bmat->i[brow]; 456 for (bcol=0; bcol<nz; bcol++){ 457 gcol = bs*garray[*jj]; jj++; 458 for (col=0; col<bs; col++){ 459 for (row=0; row<bs; row++){ 460 vabs = PetscAbsScalar(*v); v++; 461 rsum[gcol+col] += vabs; 462 rsum[grow+row] += vabs; 463 } 464 } 465 } 466 } 467 ierr = MPI_Allreduce(rsum,rsum2,mat->cmap.N,MPIU_REAL,MPI_SUM,((PetscObject)mat)->comm);CHKERRQ(ierr); 468 *norm = 0.0; 469 for (col=0; col<mat->cmap.N; col++) { 470 if (rsum2[col] > *norm) *norm = rsum2[col]; 471 } 472 ierr = PetscFree(rsum);CHKERRQ(ierr); 473 } else { 474 SETERRQ(PETSC_ERR_SUP,"No support for this norm yet"); 475 } 476 } 477 PetscFunctionReturn(0); 478 } 479 480 #undef __FUNCT__ 481 #define __FUNCT__ "MatAssemblyBegin_MPISBAIJ" 482 PetscErrorCode MatAssemblyBegin_MPISBAIJ(Mat mat,MatAssemblyType mode) 483 { 484 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 485 PetscErrorCode ierr; 486 PetscInt nstash,reallocs; 487 InsertMode addv; 488 489 PetscFunctionBegin; 490 if (baij->donotstash) { 491 PetscFunctionReturn(0); 492 } 493 494 /* make sure all processors are either in INSERTMODE or ADDMODE */ 495 ierr = MPI_Allreduce(&mat->insertmode,&addv,1,MPI_INT,MPI_BOR,((PetscObject)mat)->comm);CHKERRQ(ierr); 496 if (addv == (ADD_VALUES|INSERT_VALUES)) { 497 SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Some processors inserted others added"); 498 } 499 mat->insertmode = addv; /* in case this processor had no cache */ 500 501 ierr = MatStashScatterBegin_Private(mat,&mat->stash,mat->rmap.range);CHKERRQ(ierr); 502 ierr = MatStashScatterBegin_Private(mat,&mat->bstash,baij->rangebs);CHKERRQ(ierr); 503 ierr = MatStashGetInfo_Private(&mat->stash,&nstash,&reallocs);CHKERRQ(ierr); 504 ierr = PetscInfo2(mat,"Stash has %D entries,uses %D mallocs.\n",nstash,reallocs);CHKERRQ(ierr); 505 ierr = MatStashGetInfo_Private(&mat->stash,&nstash,&reallocs);CHKERRQ(ierr); 506 ierr = PetscInfo2(mat,"Block-Stash has %D entries, uses %D mallocs.\n",nstash,reallocs);CHKERRQ(ierr); 507 PetscFunctionReturn(0); 508 } 509 510 #undef __FUNCT__ 511 #define __FUNCT__ "MatAssemblyEnd_MPISBAIJ" 512 PetscErrorCode MatAssemblyEnd_MPISBAIJ(Mat mat,MatAssemblyType mode) 513 { 514 Mat_MPISBAIJ *baij=(Mat_MPISBAIJ*)mat->data; 515 Mat_SeqSBAIJ *a=(Mat_SeqSBAIJ*)baij->A->data; 516 PetscErrorCode ierr; 517 PetscInt i,j,rstart,ncols,flg,bs2=baij->bs2; 518 PetscInt *row,*col; 519 PetscTruth other_disassembled; 520 PetscMPIInt n; 521 PetscTruth r1,r2,r3; 522 MatScalar *val; 523 InsertMode addv = mat->insertmode; 524 525 /* do not use 'b=(Mat_SeqBAIJ*)baij->B->data' as B can be reset in disassembly */ 526 PetscFunctionBegin; 527 528 if (!baij->donotstash) { 529 while (1) { 530 ierr = MatStashScatterGetMesg_Private(&mat->stash,&n,&row,&col,&val,&flg);CHKERRQ(ierr); 531 if (!flg) break; 532 533 for (i=0; i<n;) { 534 /* Now identify the consecutive vals belonging to the same row */ 535 for (j=i,rstart=row[j]; j<n; j++) { if (row[j] != rstart) break; } 536 if (j < n) ncols = j-i; 537 else ncols = n-i; 538 /* Now assemble all these values with a single function call */ 539 ierr = MatSetValues_MPISBAIJ(mat,1,row+i,ncols,col+i,val+i,addv);CHKERRQ(ierr); 540 i = j; 541 } 542 } 543 ierr = MatStashScatterEnd_Private(&mat->stash);CHKERRQ(ierr); 544 /* Now process the block-stash. Since the values are stashed column-oriented, 545 set the roworiented flag to column oriented, and after MatSetValues() 546 restore the original flags */ 547 r1 = baij->roworiented; 548 r2 = a->roworiented; 549 r3 = ((Mat_SeqBAIJ*)baij->B->data)->roworiented; 550 baij->roworiented = PETSC_FALSE; 551 a->roworiented = PETSC_FALSE; 552 ((Mat_SeqBAIJ*)baij->B->data)->roworiented = PETSC_FALSE; /* b->roworinted */ 553 while (1) { 554 ierr = MatStashScatterGetMesg_Private(&mat->bstash,&n,&row,&col,&val,&flg);CHKERRQ(ierr); 555 if (!flg) break; 556 557 for (i=0; i<n;) { 558 /* Now identify the consecutive vals belonging to the same row */ 559 for (j=i,rstart=row[j]; j<n; j++) { if (row[j] != rstart) break; } 560 if (j < n) ncols = j-i; 561 else ncols = n-i; 562 ierr = MatSetValuesBlocked_MPISBAIJ(mat,1,row+i,ncols,col+i,val+i*bs2,addv);CHKERRQ(ierr); 563 i = j; 564 } 565 } 566 ierr = MatStashScatterEnd_Private(&mat->bstash);CHKERRQ(ierr); 567 baij->roworiented = r1; 568 a->roworiented = r2; 569 ((Mat_SeqBAIJ*)baij->B->data)->roworiented = r3; /* b->roworinted */ 570 } 571 572 ierr = MatAssemblyBegin(baij->A,mode);CHKERRQ(ierr); 573 ierr = MatAssemblyEnd(baij->A,mode);CHKERRQ(ierr); 574 575 /* determine if any processor has disassembled, if so we must 576 also disassemble ourselfs, in order that we may reassemble. */ 577 /* 578 if nonzero structure of submatrix B cannot change then we know that 579 no processor disassembled thus we can skip this stuff 580 */ 581 if (!((Mat_SeqBAIJ*)baij->B->data)->nonew) { 582 ierr = MPI_Allreduce(&mat->was_assembled,&other_disassembled,1,MPI_INT,MPI_PROD,((PetscObject)mat)->comm);CHKERRQ(ierr); 583 if (mat->was_assembled && !other_disassembled) { 584 ierr = DisAssemble_MPISBAIJ(mat);CHKERRQ(ierr); 585 } 586 } 587 588 if (!mat->was_assembled && mode == MAT_FINAL_ASSEMBLY) { 589 ierr = MatSetUpMultiply_MPISBAIJ(mat);CHKERRQ(ierr); /* setup Mvctx and sMvctx */ 590 } 591 ((Mat_SeqBAIJ*)baij->B->data)->compressedrow.use = PETSC_TRUE; /* b->compressedrow.use */ 592 ierr = MatAssemblyBegin(baij->B,mode);CHKERRQ(ierr); 593 ierr = MatAssemblyEnd(baij->B,mode);CHKERRQ(ierr); 594 595 ierr = PetscFree(baij->rowvalues);CHKERRQ(ierr); 596 baij->rowvalues = 0; 597 598 PetscFunctionReturn(0); 599 } 600 601 extern PetscErrorCode MatSetValues_MPIBAIJ(Mat,PetscInt,const PetscInt[],PetscInt,const PetscInt[],const PetscScalar[],InsertMode); 602 #undef __FUNCT__ 603 #define __FUNCT__ "MatView_MPISBAIJ_ASCIIorDraworSocket" 604 static PetscErrorCode MatView_MPISBAIJ_ASCIIorDraworSocket(Mat mat,PetscViewer viewer) 605 { 606 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 607 PetscErrorCode ierr; 608 PetscInt bs = mat->rmap.bs; 609 PetscMPIInt size = baij->size,rank = baij->rank; 610 PetscTruth iascii,isdraw; 611 PetscViewer sviewer; 612 PetscViewerFormat format; 613 614 PetscFunctionBegin; 615 ierr = PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_ASCII,&iascii);CHKERRQ(ierr); 616 ierr = PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_DRAW,&isdraw);CHKERRQ(ierr); 617 if (iascii) { 618 ierr = PetscViewerGetFormat(viewer,&format);CHKERRQ(ierr); 619 if (format == PETSC_VIEWER_ASCII_INFO_DETAIL) { 620 MatInfo info; 621 ierr = MPI_Comm_rank(((PetscObject)mat)->comm,&rank);CHKERRQ(ierr); 622 ierr = MatGetInfo(mat,MAT_LOCAL,&info);CHKERRQ(ierr); 623 ierr = PetscViewerASCIISynchronizedPrintf(viewer,"[%d] Local rows %D nz %D nz alloced %D bs %D mem %D\n", 624 rank,mat->rmap.N,(PetscInt)info.nz_used*bs,(PetscInt)info.nz_allocated*bs, 625 mat->rmap.bs,(PetscInt)info.memory);CHKERRQ(ierr); 626 ierr = MatGetInfo(baij->A,MAT_LOCAL,&info);CHKERRQ(ierr); 627 ierr = PetscViewerASCIISynchronizedPrintf(viewer,"[%d] on-diagonal part: nz %D \n",rank,(PetscInt)info.nz_used*bs);CHKERRQ(ierr); 628 ierr = MatGetInfo(baij->B,MAT_LOCAL,&info);CHKERRQ(ierr); 629 ierr = PetscViewerASCIISynchronizedPrintf(viewer,"[%d] off-diagonal part: nz %D \n",rank,(PetscInt)info.nz_used*bs);CHKERRQ(ierr); 630 ierr = PetscViewerFlush(viewer);CHKERRQ(ierr); 631 ierr = PetscViewerASCIIPrintf(viewer,"Information on VecScatter used in matrix-vector product: \n");CHKERRQ(ierr); 632 ierr = VecScatterView(baij->Mvctx,viewer);CHKERRQ(ierr); 633 PetscFunctionReturn(0); 634 } else if (format == PETSC_VIEWER_ASCII_INFO) { 635 ierr = PetscViewerASCIIPrintf(viewer," block size is %D\n",bs);CHKERRQ(ierr); 636 PetscFunctionReturn(0); 637 } else if (format == PETSC_VIEWER_ASCII_FACTOR_INFO) { 638 PetscFunctionReturn(0); 639 } 640 } 641 642 if (isdraw) { 643 PetscDraw draw; 644 PetscTruth isnull; 645 ierr = PetscViewerDrawGetDraw(viewer,0,&draw);CHKERRQ(ierr); 646 ierr = PetscDrawIsNull(draw,&isnull);CHKERRQ(ierr); if (isnull) PetscFunctionReturn(0); 647 } 648 649 if (size == 1) { 650 ierr = PetscObjectSetName((PetscObject)baij->A,((PetscObject)mat)->name);CHKERRQ(ierr); 651 ierr = MatView(baij->A,viewer);CHKERRQ(ierr); 652 } else { 653 /* assemble the entire matrix onto first processor. */ 654 Mat A; 655 Mat_SeqSBAIJ *Aloc; 656 Mat_SeqBAIJ *Bloc; 657 PetscInt M = mat->rmap.N,N = mat->cmap.N,*ai,*aj,col,i,j,k,*rvals,mbs = baij->mbs; 658 MatScalar *a; 659 660 /* Should this be the same type as mat? */ 661 ierr = MatCreate(((PetscObject)mat)->comm,&A);CHKERRQ(ierr); 662 if (!rank) { 663 ierr = MatSetSizes(A,M,N,M,N);CHKERRQ(ierr); 664 } else { 665 ierr = MatSetSizes(A,0,0,M,N);CHKERRQ(ierr); 666 } 667 ierr = MatSetType(A,MATMPISBAIJ);CHKERRQ(ierr); 668 ierr = MatMPISBAIJSetPreallocation(A,mat->rmap.bs,0,PETSC_NULL,0,PETSC_NULL);CHKERRQ(ierr); 669 ierr = PetscLogObjectParent(mat,A);CHKERRQ(ierr); 670 671 /* copy over the A part */ 672 Aloc = (Mat_SeqSBAIJ*)baij->A->data; 673 ai = Aloc->i; aj = Aloc->j; a = Aloc->a; 674 ierr = PetscMalloc(bs*sizeof(PetscInt),&rvals);CHKERRQ(ierr); 675 676 for (i=0; i<mbs; i++) { 677 rvals[0] = bs*(baij->rstartbs + i); 678 for (j=1; j<bs; j++) { rvals[j] = rvals[j-1] + 1; } 679 for (j=ai[i]; j<ai[i+1]; j++) { 680 col = (baij->cstartbs+aj[j])*bs; 681 for (k=0; k<bs; k++) { 682 ierr = MatSetValues_MPISBAIJ(A,bs,rvals,1,&col,a,INSERT_VALUES);CHKERRQ(ierr); 683 col++; a += bs; 684 } 685 } 686 } 687 /* copy over the B part */ 688 Bloc = (Mat_SeqBAIJ*)baij->B->data; 689 ai = Bloc->i; aj = Bloc->j; a = Bloc->a; 690 for (i=0; i<mbs; i++) { 691 692 rvals[0] = bs*(baij->rstartbs + i); 693 for (j=1; j<bs; j++) { rvals[j] = rvals[j-1] + 1; } 694 for (j=ai[i]; j<ai[i+1]; j++) { 695 col = baij->garray[aj[j]]*bs; 696 for (k=0; k<bs; k++) { 697 ierr = MatSetValues_MPIBAIJ(A,bs,rvals,1,&col,a,INSERT_VALUES);CHKERRQ(ierr); 698 col++; a += bs; 699 } 700 } 701 } 702 ierr = PetscFree(rvals);CHKERRQ(ierr); 703 ierr = MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 704 ierr = MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 705 /* 706 Everyone has to call to draw the matrix since the graphics waits are 707 synchronized across all processors that share the PetscDraw object 708 */ 709 ierr = PetscViewerGetSingleton(viewer,&sviewer);CHKERRQ(ierr); 710 if (!rank) { 711 ierr = PetscObjectSetName((PetscObject)((Mat_MPISBAIJ*)(A->data))->A,((PetscObject)mat)->name);CHKERRQ(ierr); 712 ierr = MatView(((Mat_MPISBAIJ*)(A->data))->A,sviewer);CHKERRQ(ierr); 713 } 714 ierr = PetscViewerRestoreSingleton(viewer,&sviewer);CHKERRQ(ierr); 715 ierr = MatDestroy(A);CHKERRQ(ierr); 716 } 717 PetscFunctionReturn(0); 718 } 719 720 #undef __FUNCT__ 721 #define __FUNCT__ "MatView_MPISBAIJ" 722 PetscErrorCode MatView_MPISBAIJ(Mat mat,PetscViewer viewer) 723 { 724 PetscErrorCode ierr; 725 PetscTruth iascii,isdraw,issocket,isbinary; 726 727 PetscFunctionBegin; 728 ierr = PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_ASCII,&iascii);CHKERRQ(ierr); 729 ierr = PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_DRAW,&isdraw);CHKERRQ(ierr); 730 ierr = PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_SOCKET,&issocket);CHKERRQ(ierr); 731 ierr = PetscTypeCompare((PetscObject)viewer,PETSC_VIEWER_BINARY,&isbinary);CHKERRQ(ierr); 732 if (iascii || isdraw || issocket || isbinary) { 733 ierr = MatView_MPISBAIJ_ASCIIorDraworSocket(mat,viewer);CHKERRQ(ierr); 734 } else { 735 SETERRQ1(PETSC_ERR_SUP,"Viewer type %s not supported by MPISBAIJ matrices",((PetscObject)viewer)->type_name); 736 } 737 PetscFunctionReturn(0); 738 } 739 740 #undef __FUNCT__ 741 #define __FUNCT__ "MatDestroy_MPISBAIJ" 742 PetscErrorCode MatDestroy_MPISBAIJ(Mat mat) 743 { 744 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 745 PetscErrorCode ierr; 746 747 PetscFunctionBegin; 748 #if defined(PETSC_USE_LOG) 749 PetscLogObjectState((PetscObject)mat,"Rows=%D,Cols=%D",mat->rmap.N,mat->cmap.N); 750 #endif 751 ierr = MatStashDestroy_Private(&mat->stash);CHKERRQ(ierr); 752 ierr = MatStashDestroy_Private(&mat->bstash);CHKERRQ(ierr); 753 ierr = MatDestroy(baij->A);CHKERRQ(ierr); 754 ierr = MatDestroy(baij->B);CHKERRQ(ierr); 755 #if defined (PETSC_USE_CTABLE) 756 if (baij->colmap) {ierr = PetscTableDestroy(baij->colmap);CHKERRQ(ierr);} 757 #else 758 ierr = PetscFree(baij->colmap);CHKERRQ(ierr); 759 #endif 760 ierr = PetscFree(baij->garray);CHKERRQ(ierr); 761 if (baij->lvec) {ierr = VecDestroy(baij->lvec);CHKERRQ(ierr);} 762 if (baij->Mvctx) {ierr = VecScatterDestroy(baij->Mvctx);CHKERRQ(ierr);} 763 if (baij->slvec0) { 764 ierr = VecDestroy(baij->slvec0);CHKERRQ(ierr); 765 ierr = VecDestroy(baij->slvec0b);CHKERRQ(ierr); 766 } 767 if (baij->slvec1) { 768 ierr = VecDestroy(baij->slvec1);CHKERRQ(ierr); 769 ierr = VecDestroy(baij->slvec1a);CHKERRQ(ierr); 770 ierr = VecDestroy(baij->slvec1b);CHKERRQ(ierr); 771 } 772 if (baij->sMvctx) {ierr = VecScatterDestroy(baij->sMvctx);CHKERRQ(ierr);} 773 ierr = PetscFree(baij->rowvalues);CHKERRQ(ierr); 774 ierr = PetscFree(baij->barray);CHKERRQ(ierr); 775 ierr = PetscFree(baij->hd);CHKERRQ(ierr); 776 #if defined(PETSC_USE_MAT_SINGLE) 777 ierr = PetscFree(baij->setvaluescopy);CHKERRQ(ierr); 778 #endif 779 ierr = PetscFree(baij->in_loc);CHKERRQ(ierr); 780 ierr = PetscFree(baij->v_loc);CHKERRQ(ierr); 781 ierr = PetscFree(baij->rangebs);CHKERRQ(ierr); 782 ierr = PetscFree(baij);CHKERRQ(ierr); 783 784 ierr = PetscObjectChangeTypeName((PetscObject)mat,0);CHKERRQ(ierr); 785 ierr = PetscObjectComposeFunction((PetscObject)mat,"MatStoreValues_C","",PETSC_NULL);CHKERRQ(ierr); 786 ierr = PetscObjectComposeFunction((PetscObject)mat,"MatRetrieveValues_C","",PETSC_NULL);CHKERRQ(ierr); 787 ierr = PetscObjectComposeFunction((PetscObject)mat,"MatGetDiagonalBlock_C","",PETSC_NULL);CHKERRQ(ierr); 788 ierr = PetscObjectComposeFunction((PetscObject)mat,"MatMPISBAIJSetPreallocation_C","",PETSC_NULL);CHKERRQ(ierr); 789 PetscFunctionReturn(0); 790 } 791 792 #undef __FUNCT__ 793 #define __FUNCT__ "MatMult_MPISBAIJ" 794 PetscErrorCode MatMult_MPISBAIJ(Mat A,Vec xx,Vec yy) 795 { 796 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 797 PetscErrorCode ierr; 798 PetscInt nt,mbs=a->mbs,bs=A->rmap.bs; 799 PetscScalar *x,*from,zero=0.0; 800 801 PetscFunctionBegin; 802 ierr = VecGetLocalSize(xx,&nt);CHKERRQ(ierr); 803 if (nt != A->cmap.n) { 804 SETERRQ(PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx"); 805 } 806 807 /* diagonal part */ 808 ierr = (*a->A->ops->mult)(a->A,xx,a->slvec1a);CHKERRQ(ierr); 809 ierr = VecSet(a->slvec1b,zero);CHKERRQ(ierr); 810 811 /* subdiagonal part */ 812 ierr = (*a->B->ops->multtranspose)(a->B,xx,a->slvec0b);CHKERRQ(ierr); 813 814 /* copy x into the vec slvec0 */ 815 ierr = VecGetArray(a->slvec0,&from);CHKERRQ(ierr); 816 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 817 818 ierr = PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));CHKERRQ(ierr); 819 ierr = VecRestoreArray(a->slvec0,&from);CHKERRQ(ierr); 820 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 821 822 ierr = VecScatterBegin(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 823 ierr = VecScatterEnd(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 824 /* supperdiagonal part */ 825 ierr = (*a->B->ops->multadd)(a->B,a->slvec1b,a->slvec1a,yy);CHKERRQ(ierr); 826 PetscFunctionReturn(0); 827 } 828 829 #undef __FUNCT__ 830 #define __FUNCT__ "MatMult_MPISBAIJ_2comm" 831 PetscErrorCode MatMult_MPISBAIJ_2comm(Mat A,Vec xx,Vec yy) 832 { 833 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 834 PetscErrorCode ierr; 835 PetscInt nt; 836 837 PetscFunctionBegin; 838 ierr = VecGetLocalSize(xx,&nt);CHKERRQ(ierr); 839 if (nt != A->cmap.n) { 840 SETERRQ(PETSC_ERR_ARG_SIZ,"Incompatible partition of A and xx"); 841 } 842 ierr = VecGetLocalSize(yy,&nt);CHKERRQ(ierr); 843 if (nt != A->rmap.N) { 844 SETERRQ(PETSC_ERR_ARG_SIZ,"Incompatible parition of A and yy"); 845 } 846 847 ierr = VecScatterBegin(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 848 /* do diagonal part */ 849 ierr = (*a->A->ops->mult)(a->A,xx,yy);CHKERRQ(ierr); 850 /* do supperdiagonal part */ 851 ierr = VecScatterEnd(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 852 ierr = (*a->B->ops->multadd)(a->B,a->lvec,yy,yy);CHKERRQ(ierr); 853 /* do subdiagonal part */ 854 ierr = (*a->B->ops->multtranspose)(a->B,xx,a->lvec);CHKERRQ(ierr); 855 ierr = VecScatterBegin(a->Mvctx,a->lvec,yy,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 856 ierr = VecScatterEnd(a->Mvctx,a->lvec,yy,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 857 858 PetscFunctionReturn(0); 859 } 860 861 #undef __FUNCT__ 862 #define __FUNCT__ "MatMultAdd_MPISBAIJ" 863 PetscErrorCode MatMultAdd_MPISBAIJ(Mat A,Vec xx,Vec yy,Vec zz) 864 { 865 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 866 PetscErrorCode ierr; 867 PetscInt mbs=a->mbs,bs=A->rmap.bs; 868 PetscScalar *x,*from,zero=0.0; 869 870 PetscFunctionBegin; 871 /* 872 PetscSynchronizedPrintf(((PetscObject)A)->comm," MatMultAdd is called ...\n"); 873 PetscSynchronizedFlush(((PetscObject)A)->comm); 874 */ 875 /* diagonal part */ 876 ierr = (*a->A->ops->multadd)(a->A,xx,yy,a->slvec1a);CHKERRQ(ierr); 877 ierr = VecSet(a->slvec1b,zero);CHKERRQ(ierr); 878 879 /* subdiagonal part */ 880 ierr = (*a->B->ops->multtranspose)(a->B,xx,a->slvec0b);CHKERRQ(ierr); 881 882 /* copy x into the vec slvec0 */ 883 ierr = VecGetArray(a->slvec0,&from);CHKERRQ(ierr); 884 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 885 ierr = PetscMemcpy(from,x,bs*mbs*sizeof(MatScalar));CHKERRQ(ierr); 886 ierr = VecRestoreArray(a->slvec0,&from);CHKERRQ(ierr); 887 888 ierr = VecScatterBegin(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 889 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 890 ierr = VecScatterEnd(a->sMvctx,a->slvec0,a->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 891 892 /* supperdiagonal part */ 893 ierr = (*a->B->ops->multadd)(a->B,a->slvec1b,a->slvec1a,zz);CHKERRQ(ierr); 894 895 PetscFunctionReturn(0); 896 } 897 898 #undef __FUNCT__ 899 #define __FUNCT__ "MatMultAdd_MPISBAIJ_2comm" 900 PetscErrorCode MatMultAdd_MPISBAIJ_2comm(Mat A,Vec xx,Vec yy,Vec zz) 901 { 902 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 903 PetscErrorCode ierr; 904 905 PetscFunctionBegin; 906 ierr = VecScatterBegin(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 907 /* do diagonal part */ 908 ierr = (*a->A->ops->multadd)(a->A,xx,yy,zz);CHKERRQ(ierr); 909 /* do supperdiagonal part */ 910 ierr = VecScatterEnd(a->Mvctx,xx,a->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 911 ierr = (*a->B->ops->multadd)(a->B,a->lvec,zz,zz);CHKERRQ(ierr); 912 913 /* do subdiagonal part */ 914 ierr = (*a->B->ops->multtranspose)(a->B,xx,a->lvec);CHKERRQ(ierr); 915 ierr = VecScatterBegin(a->Mvctx,a->lvec,zz,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 916 ierr = VecScatterEnd(a->Mvctx,a->lvec,zz,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 917 918 PetscFunctionReturn(0); 919 } 920 921 /* 922 This only works correctly for square matrices where the subblock A->A is the 923 diagonal block 924 */ 925 #undef __FUNCT__ 926 #define __FUNCT__ "MatGetDiagonal_MPISBAIJ" 927 PetscErrorCode MatGetDiagonal_MPISBAIJ(Mat A,Vec v) 928 { 929 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 930 PetscErrorCode ierr; 931 932 PetscFunctionBegin; 933 /* if (a->rmap.N != a->cmap.N) SETERRQ(PETSC_ERR_SUP,"Supports only square matrix where A->A is diag block"); */ 934 ierr = MatGetDiagonal(a->A,v);CHKERRQ(ierr); 935 PetscFunctionReturn(0); 936 } 937 938 #undef __FUNCT__ 939 #define __FUNCT__ "MatScale_MPISBAIJ" 940 PetscErrorCode MatScale_MPISBAIJ(Mat A,PetscScalar aa) 941 { 942 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 943 PetscErrorCode ierr; 944 945 PetscFunctionBegin; 946 ierr = MatScale(a->A,aa);CHKERRQ(ierr); 947 ierr = MatScale(a->B,aa);CHKERRQ(ierr); 948 PetscFunctionReturn(0); 949 } 950 951 #undef __FUNCT__ 952 #define __FUNCT__ "MatGetRow_MPISBAIJ" 953 PetscErrorCode MatGetRow_MPISBAIJ(Mat matin,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v) 954 { 955 Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data; 956 PetscScalar *vworkA,*vworkB,**pvA,**pvB,*v_p; 957 PetscErrorCode ierr; 958 PetscInt bs = matin->rmap.bs,bs2 = mat->bs2,i,*cworkA,*cworkB,**pcA,**pcB; 959 PetscInt nztot,nzA,nzB,lrow,brstart = matin->rmap.rstart,brend = matin->rmap.rend; 960 PetscInt *cmap,*idx_p,cstart = mat->rstartbs; 961 962 PetscFunctionBegin; 963 if (mat->getrowactive == PETSC_TRUE) SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"Already active"); 964 mat->getrowactive = PETSC_TRUE; 965 966 if (!mat->rowvalues && (idx || v)) { 967 /* 968 allocate enough space to hold information from the longest row. 969 */ 970 Mat_SeqSBAIJ *Aa = (Mat_SeqSBAIJ*)mat->A->data; 971 Mat_SeqBAIJ *Ba = (Mat_SeqBAIJ*)mat->B->data; 972 PetscInt max = 1,mbs = mat->mbs,tmp; 973 for (i=0; i<mbs; i++) { 974 tmp = Aa->i[i+1] - Aa->i[i] + Ba->i[i+1] - Ba->i[i]; /* row length */ 975 if (max < tmp) { max = tmp; } 976 } 977 ierr = PetscMalloc(max*bs2*(sizeof(PetscInt)+sizeof(PetscScalar)),&mat->rowvalues);CHKERRQ(ierr); 978 mat->rowindices = (PetscInt*)(mat->rowvalues + max*bs2); 979 } 980 981 if (row < brstart || row >= brend) SETERRQ(PETSC_ERR_SUP,"Only local rows") 982 lrow = row - brstart; /* local row index */ 983 984 pvA = &vworkA; pcA = &cworkA; pvB = &vworkB; pcB = &cworkB; 985 if (!v) {pvA = 0; pvB = 0;} 986 if (!idx) {pcA = 0; if (!v) pcB = 0;} 987 ierr = (*mat->A->ops->getrow)(mat->A,lrow,&nzA,pcA,pvA);CHKERRQ(ierr); 988 ierr = (*mat->B->ops->getrow)(mat->B,lrow,&nzB,pcB,pvB);CHKERRQ(ierr); 989 nztot = nzA + nzB; 990 991 cmap = mat->garray; 992 if (v || idx) { 993 if (nztot) { 994 /* Sort by increasing column numbers, assuming A and B already sorted */ 995 PetscInt imark = -1; 996 if (v) { 997 *v = v_p = mat->rowvalues; 998 for (i=0; i<nzB; i++) { 999 if (cmap[cworkB[i]/bs] < cstart) v_p[i] = vworkB[i]; 1000 else break; 1001 } 1002 imark = i; 1003 for (i=0; i<nzA; i++) v_p[imark+i] = vworkA[i]; 1004 for (i=imark; i<nzB; i++) v_p[nzA+i] = vworkB[i]; 1005 } 1006 if (idx) { 1007 *idx = idx_p = mat->rowindices; 1008 if (imark > -1) { 1009 for (i=0; i<imark; i++) { 1010 idx_p[i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs; 1011 } 1012 } else { 1013 for (i=0; i<nzB; i++) { 1014 if (cmap[cworkB[i]/bs] < cstart) 1015 idx_p[i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs ; 1016 else break; 1017 } 1018 imark = i; 1019 } 1020 for (i=0; i<nzA; i++) idx_p[imark+i] = cstart*bs + cworkA[i]; 1021 for (i=imark; i<nzB; i++) idx_p[nzA+i] = cmap[cworkB[i]/bs]*bs + cworkB[i]%bs ; 1022 } 1023 } else { 1024 if (idx) *idx = 0; 1025 if (v) *v = 0; 1026 } 1027 } 1028 *nz = nztot; 1029 ierr = (*mat->A->ops->restorerow)(mat->A,lrow,&nzA,pcA,pvA);CHKERRQ(ierr); 1030 ierr = (*mat->B->ops->restorerow)(mat->B,lrow,&nzB,pcB,pvB);CHKERRQ(ierr); 1031 PetscFunctionReturn(0); 1032 } 1033 1034 #undef __FUNCT__ 1035 #define __FUNCT__ "MatRestoreRow_MPISBAIJ" 1036 PetscErrorCode MatRestoreRow_MPISBAIJ(Mat mat,PetscInt row,PetscInt *nz,PetscInt **idx,PetscScalar **v) 1037 { 1038 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 1039 1040 PetscFunctionBegin; 1041 if (!baij->getrowactive) { 1042 SETERRQ(PETSC_ERR_ARG_WRONGSTATE,"MatGetRow() must be called first"); 1043 } 1044 baij->getrowactive = PETSC_FALSE; 1045 PetscFunctionReturn(0); 1046 } 1047 1048 #undef __FUNCT__ 1049 #define __FUNCT__ "MatGetRowUpperTriangular_MPISBAIJ" 1050 PetscErrorCode MatGetRowUpperTriangular_MPISBAIJ(Mat A) 1051 { 1052 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1053 Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data; 1054 1055 PetscFunctionBegin; 1056 aA->getrow_utriangular = PETSC_TRUE; 1057 PetscFunctionReturn(0); 1058 } 1059 #undef __FUNCT__ 1060 #define __FUNCT__ "MatRestoreRowUpperTriangular_MPISBAIJ" 1061 PetscErrorCode MatRestoreRowUpperTriangular_MPISBAIJ(Mat A) 1062 { 1063 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1064 Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data; 1065 1066 PetscFunctionBegin; 1067 aA->getrow_utriangular = PETSC_FALSE; 1068 PetscFunctionReturn(0); 1069 } 1070 1071 #undef __FUNCT__ 1072 #define __FUNCT__ "MatRealPart_MPISBAIJ" 1073 PetscErrorCode MatRealPart_MPISBAIJ(Mat A) 1074 { 1075 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1076 PetscErrorCode ierr; 1077 1078 PetscFunctionBegin; 1079 ierr = MatRealPart(a->A);CHKERRQ(ierr); 1080 ierr = MatRealPart(a->B);CHKERRQ(ierr); 1081 PetscFunctionReturn(0); 1082 } 1083 1084 #undef __FUNCT__ 1085 #define __FUNCT__ "MatImaginaryPart_MPISBAIJ" 1086 PetscErrorCode MatImaginaryPart_MPISBAIJ(Mat A) 1087 { 1088 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1089 PetscErrorCode ierr; 1090 1091 PetscFunctionBegin; 1092 ierr = MatImaginaryPart(a->A);CHKERRQ(ierr); 1093 ierr = MatImaginaryPart(a->B);CHKERRQ(ierr); 1094 PetscFunctionReturn(0); 1095 } 1096 1097 #undef __FUNCT__ 1098 #define __FUNCT__ "MatZeroEntries_MPISBAIJ" 1099 PetscErrorCode MatZeroEntries_MPISBAIJ(Mat A) 1100 { 1101 Mat_MPISBAIJ *l = (Mat_MPISBAIJ*)A->data; 1102 PetscErrorCode ierr; 1103 1104 PetscFunctionBegin; 1105 ierr = MatZeroEntries(l->A);CHKERRQ(ierr); 1106 ierr = MatZeroEntries(l->B);CHKERRQ(ierr); 1107 PetscFunctionReturn(0); 1108 } 1109 1110 #undef __FUNCT__ 1111 #define __FUNCT__ "MatGetInfo_MPISBAIJ" 1112 PetscErrorCode MatGetInfo_MPISBAIJ(Mat matin,MatInfoType flag,MatInfo *info) 1113 { 1114 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)matin->data; 1115 Mat A = a->A,B = a->B; 1116 PetscErrorCode ierr; 1117 PetscReal isend[5],irecv[5]; 1118 1119 PetscFunctionBegin; 1120 info->block_size = (PetscReal)matin->rmap.bs; 1121 ierr = MatGetInfo(A,MAT_LOCAL,info);CHKERRQ(ierr); 1122 isend[0] = info->nz_used; isend[1] = info->nz_allocated; isend[2] = info->nz_unneeded; 1123 isend[3] = info->memory; isend[4] = info->mallocs; 1124 ierr = MatGetInfo(B,MAT_LOCAL,info);CHKERRQ(ierr); 1125 isend[0] += info->nz_used; isend[1] += info->nz_allocated; isend[2] += info->nz_unneeded; 1126 isend[3] += info->memory; isend[4] += info->mallocs; 1127 if (flag == MAT_LOCAL) { 1128 info->nz_used = isend[0]; 1129 info->nz_allocated = isend[1]; 1130 info->nz_unneeded = isend[2]; 1131 info->memory = isend[3]; 1132 info->mallocs = isend[4]; 1133 } else if (flag == MAT_GLOBAL_MAX) { 1134 ierr = MPI_Allreduce(isend,irecv,5,MPIU_REAL,MPI_MAX,((PetscObject)matin)->comm);CHKERRQ(ierr); 1135 info->nz_used = irecv[0]; 1136 info->nz_allocated = irecv[1]; 1137 info->nz_unneeded = irecv[2]; 1138 info->memory = irecv[3]; 1139 info->mallocs = irecv[4]; 1140 } else if (flag == MAT_GLOBAL_SUM) { 1141 ierr = MPI_Allreduce(isend,irecv,5,MPIU_REAL,MPI_SUM,((PetscObject)matin)->comm);CHKERRQ(ierr); 1142 info->nz_used = irecv[0]; 1143 info->nz_allocated = irecv[1]; 1144 info->nz_unneeded = irecv[2]; 1145 info->memory = irecv[3]; 1146 info->mallocs = irecv[4]; 1147 } else { 1148 SETERRQ1(PETSC_ERR_ARG_WRONG,"Unknown MatInfoType argument %d",(int)flag); 1149 } 1150 info->rows_global = (PetscReal)A->rmap.N; 1151 info->columns_global = (PetscReal)A->cmap.N; 1152 info->rows_local = (PetscReal)A->rmap.N; 1153 info->columns_local = (PetscReal)A->cmap.N; 1154 info->fill_ratio_given = 0; /* no parallel LU/ILU/Cholesky */ 1155 info->fill_ratio_needed = 0; 1156 info->factor_mallocs = 0; 1157 PetscFunctionReturn(0); 1158 } 1159 1160 #undef __FUNCT__ 1161 #define __FUNCT__ "MatSetOption_MPISBAIJ" 1162 PetscErrorCode MatSetOption_MPISBAIJ(Mat A,MatOption op,PetscTruth flg) 1163 { 1164 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1165 Mat_SeqSBAIJ *aA = (Mat_SeqSBAIJ*)a->A->data; 1166 PetscErrorCode ierr; 1167 1168 PetscFunctionBegin; 1169 switch (op) { 1170 case MAT_NEW_NONZERO_LOCATIONS: 1171 case MAT_NEW_NONZERO_ALLOCATION_ERR: 1172 case MAT_KEEP_ZEROED_ROWS: 1173 case MAT_NEW_NONZERO_LOCATION_ERR: 1174 ierr = MatSetOption(a->A,op,flg);CHKERRQ(ierr); 1175 ierr = MatSetOption(a->B,op,flg);CHKERRQ(ierr); 1176 break; 1177 case MAT_ROW_ORIENTED: 1178 a->roworiented = flg; 1179 ierr = MatSetOption(a->A,op,flg);CHKERRQ(ierr); 1180 ierr = MatSetOption(a->B,op,flg);CHKERRQ(ierr); 1181 break; 1182 case MAT_NEW_DIAGONALS: 1183 ierr = PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);CHKERRQ(ierr); 1184 break; 1185 case MAT_IGNORE_OFF_PROC_ENTRIES: 1186 a->donotstash = flg; 1187 break; 1188 case MAT_USE_HASH_TABLE: 1189 a->ht_flag = flg; 1190 break; 1191 case MAT_HERMITIAN: 1192 if (flg) SETERRQ(PETSC_ERR_SUP,"Matrix must be symmetric"); 1193 case MAT_SYMMETRIC: 1194 case MAT_STRUCTURALLY_SYMMETRIC: 1195 case MAT_SYMMETRY_ETERNAL: 1196 if (!flg) SETERRQ(PETSC_ERR_SUP,"Matrix must be symmetric"); 1197 ierr = PetscInfo1(A,"Option %s ignored\n",MatOptions[op]);CHKERRQ(ierr); 1198 break; 1199 case MAT_IGNORE_LOWER_TRIANGULAR: 1200 aA->ignore_ltriangular = flg; 1201 break; 1202 case MAT_ERROR_LOWER_TRIANGULAR: 1203 aA->ignore_ltriangular = flg; 1204 break; 1205 case MAT_GETROW_UPPERTRIANGULAR: 1206 aA->getrow_utriangular = flg; 1207 break; 1208 default: 1209 SETERRQ1(PETSC_ERR_SUP,"unknown option %d",op); 1210 } 1211 PetscFunctionReturn(0); 1212 } 1213 1214 #undef __FUNCT__ 1215 #define __FUNCT__ "MatTranspose_MPISBAIJ" 1216 PetscErrorCode MatTranspose_MPISBAIJ(Mat A,MatReuse reuse,Mat *B) 1217 { 1218 PetscErrorCode ierr; 1219 PetscFunctionBegin; 1220 if (MAT_INITIAL_MATRIX || *B != A) { 1221 ierr = MatDuplicate(A,MAT_COPY_VALUES,B);CHKERRQ(ierr); 1222 } 1223 PetscFunctionReturn(0); 1224 } 1225 1226 #undef __FUNCT__ 1227 #define __FUNCT__ "MatDiagonalScale_MPISBAIJ" 1228 PetscErrorCode MatDiagonalScale_MPISBAIJ(Mat mat,Vec ll,Vec rr) 1229 { 1230 Mat_MPISBAIJ *baij = (Mat_MPISBAIJ*)mat->data; 1231 Mat a=baij->A, b=baij->B; 1232 PetscErrorCode ierr; 1233 PetscInt nv,m,n; 1234 PetscTruth flg; 1235 1236 PetscFunctionBegin; 1237 if (ll != rr){ 1238 ierr = VecEqual(ll,rr,&flg);CHKERRQ(ierr); 1239 if (!flg) 1240 SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,"For symmetric format, left and right scaling vectors must be same\n"); 1241 } 1242 if (!ll) PetscFunctionReturn(0); 1243 1244 ierr = MatGetLocalSize(mat,&m,&n);CHKERRQ(ierr); 1245 if (m != n) SETERRQ2(PETSC_ERR_ARG_SIZ,"For symmetric format, local size %d %d must be same",m,n); 1246 1247 ierr = VecGetLocalSize(rr,&nv);CHKERRQ(ierr); 1248 if (nv!=n) SETERRQ(PETSC_ERR_ARG_SIZ,"Left and right vector non-conforming local size"); 1249 1250 ierr = VecScatterBegin(baij->Mvctx,rr,baij->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 1251 1252 /* left diagonalscale the off-diagonal part */ 1253 ierr = (*b->ops->diagonalscale)(b,ll,PETSC_NULL);CHKERRQ(ierr); 1254 1255 /* scale the diagonal part */ 1256 ierr = (*a->ops->diagonalscale)(a,ll,rr);CHKERRQ(ierr); 1257 1258 /* right diagonalscale the off-diagonal part */ 1259 ierr = VecScatterEnd(baij->Mvctx,rr,baij->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 1260 ierr = (*b->ops->diagonalscale)(b,PETSC_NULL,baij->lvec);CHKERRQ(ierr); 1261 PetscFunctionReturn(0); 1262 } 1263 1264 #undef __FUNCT__ 1265 #define __FUNCT__ "MatSetUnfactored_MPISBAIJ" 1266 PetscErrorCode MatSetUnfactored_MPISBAIJ(Mat A) 1267 { 1268 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 1269 PetscErrorCode ierr; 1270 1271 PetscFunctionBegin; 1272 ierr = MatSetUnfactored(a->A);CHKERRQ(ierr); 1273 PetscFunctionReturn(0); 1274 } 1275 1276 static PetscErrorCode MatDuplicate_MPISBAIJ(Mat,MatDuplicateOption,Mat *); 1277 1278 #undef __FUNCT__ 1279 #define __FUNCT__ "MatEqual_MPISBAIJ" 1280 PetscErrorCode MatEqual_MPISBAIJ(Mat A,Mat B,PetscTruth *flag) 1281 { 1282 Mat_MPISBAIJ *matB = (Mat_MPISBAIJ*)B->data,*matA = (Mat_MPISBAIJ*)A->data; 1283 Mat a,b,c,d; 1284 PetscTruth flg; 1285 PetscErrorCode ierr; 1286 1287 PetscFunctionBegin; 1288 a = matA->A; b = matA->B; 1289 c = matB->A; d = matB->B; 1290 1291 ierr = MatEqual(a,c,&flg);CHKERRQ(ierr); 1292 if (flg) { 1293 ierr = MatEqual(b,d,&flg);CHKERRQ(ierr); 1294 } 1295 ierr = MPI_Allreduce(&flg,flag,1,MPI_INT,MPI_LAND,((PetscObject)A)->comm);CHKERRQ(ierr); 1296 PetscFunctionReturn(0); 1297 } 1298 1299 #undef __FUNCT__ 1300 #define __FUNCT__ "MatCopy_MPISBAIJ" 1301 PetscErrorCode MatCopy_MPISBAIJ(Mat A,Mat B,MatStructure str) 1302 { 1303 PetscErrorCode ierr; 1304 Mat_MPISBAIJ *a = (Mat_MPISBAIJ *)A->data; 1305 Mat_MPISBAIJ *b = (Mat_MPISBAIJ *)B->data; 1306 1307 PetscFunctionBegin; 1308 /* If the two matrices don't have the same copy implementation, they aren't compatible for fast copy. */ 1309 if ((str != SAME_NONZERO_PATTERN) || (A->ops->copy != B->ops->copy)) { 1310 ierr = MatGetRowUpperTriangular(A);CHKERRQ(ierr); 1311 ierr = MatCopy_Basic(A,B,str);CHKERRQ(ierr); 1312 ierr = MatRestoreRowUpperTriangular(A);CHKERRQ(ierr); 1313 } else { 1314 ierr = MatCopy(a->A,b->A,str);CHKERRQ(ierr); 1315 ierr = MatCopy(a->B,b->B,str);CHKERRQ(ierr); 1316 } 1317 PetscFunctionReturn(0); 1318 } 1319 1320 #undef __FUNCT__ 1321 #define __FUNCT__ "MatSetUpPreallocation_MPISBAIJ" 1322 PetscErrorCode MatSetUpPreallocation_MPISBAIJ(Mat A) 1323 { 1324 PetscErrorCode ierr; 1325 1326 PetscFunctionBegin; 1327 ierr = MatMPISBAIJSetPreallocation(A,PetscMax(A->rmap.bs,1),PETSC_DEFAULT,0,PETSC_DEFAULT,0);CHKERRQ(ierr); 1328 PetscFunctionReturn(0); 1329 } 1330 1331 #include "petscblaslapack.h" 1332 #undef __FUNCT__ 1333 #define __FUNCT__ "MatAXPY_MPISBAIJ" 1334 PetscErrorCode MatAXPY_MPISBAIJ(Mat Y,PetscScalar a,Mat X,MatStructure str) 1335 { 1336 PetscErrorCode ierr; 1337 Mat_MPISBAIJ *xx=(Mat_MPISBAIJ *)X->data,*yy=(Mat_MPISBAIJ *)Y->data; 1338 PetscBLASInt bnz,one=1; 1339 Mat_SeqSBAIJ *xa,*ya; 1340 Mat_SeqBAIJ *xb,*yb; 1341 1342 PetscFunctionBegin; 1343 if (str == SAME_NONZERO_PATTERN) { 1344 PetscScalar alpha = a; 1345 xa = (Mat_SeqSBAIJ *)xx->A->data; 1346 ya = (Mat_SeqSBAIJ *)yy->A->data; 1347 bnz = PetscBLASIntCast(xa->nz); 1348 BLASaxpy_(&bnz,&alpha,xa->a,&one,ya->a,&one); 1349 xb = (Mat_SeqBAIJ *)xx->B->data; 1350 yb = (Mat_SeqBAIJ *)yy->B->data; 1351 bnz = PetscBLASIntCast(xb->nz); 1352 BLASaxpy_(&bnz,&alpha,xb->a,&one,yb->a,&one); 1353 } else { 1354 ierr = MatGetRowUpperTriangular(X);CHKERRQ(ierr); 1355 ierr = MatAXPY_Basic(Y,a,X,str);CHKERRQ(ierr); 1356 ierr = MatRestoreRowUpperTriangular(X);CHKERRQ(ierr); 1357 } 1358 PetscFunctionReturn(0); 1359 } 1360 1361 #undef __FUNCT__ 1362 #define __FUNCT__ "MatGetSubMatrices_MPISBAIJ" 1363 PetscErrorCode MatGetSubMatrices_MPISBAIJ(Mat A,PetscInt n,const IS irow[],const IS icol[],MatReuse scall,Mat *B[]) 1364 { 1365 PetscErrorCode ierr; 1366 PetscInt i; 1367 PetscTruth flg; 1368 1369 PetscFunctionBegin; 1370 for (i=0; i<n; i++) { 1371 ierr = ISEqual(irow[i],icol[i],&flg);CHKERRQ(ierr); 1372 if (!flg) { 1373 SETERRQ(PETSC_ERR_SUP,"Can only get symmetric submatrix for MPISBAIJ matrices"); 1374 } 1375 } 1376 ierr = MatGetSubMatrices_MPIBAIJ(A,n,irow,icol,scall,B);CHKERRQ(ierr); 1377 PetscFunctionReturn(0); 1378 } 1379 1380 1381 /* -------------------------------------------------------------------*/ 1382 static struct _MatOps MatOps_Values = { 1383 MatSetValues_MPISBAIJ, 1384 MatGetRow_MPISBAIJ, 1385 MatRestoreRow_MPISBAIJ, 1386 MatMult_MPISBAIJ, 1387 /* 4*/ MatMultAdd_MPISBAIJ, 1388 MatMult_MPISBAIJ, /* transpose versions are same as non-transpose */ 1389 MatMultAdd_MPISBAIJ, 1390 0, 1391 0, 1392 0, 1393 /*10*/ 0, 1394 0, 1395 0, 1396 MatRelax_MPISBAIJ, 1397 MatTranspose_MPISBAIJ, 1398 /*15*/ MatGetInfo_MPISBAIJ, 1399 MatEqual_MPISBAIJ, 1400 MatGetDiagonal_MPISBAIJ, 1401 MatDiagonalScale_MPISBAIJ, 1402 MatNorm_MPISBAIJ, 1403 /*20*/ MatAssemblyBegin_MPISBAIJ, 1404 MatAssemblyEnd_MPISBAIJ, 1405 0, 1406 MatSetOption_MPISBAIJ, 1407 MatZeroEntries_MPISBAIJ, 1408 /*25*/ 0, 1409 0, 1410 0, 1411 0, 1412 0, 1413 /*30*/ MatSetUpPreallocation_MPISBAIJ, 1414 0, 1415 0, 1416 0, 1417 0, 1418 /*35*/ MatDuplicate_MPISBAIJ, 1419 0, 1420 0, 1421 0, 1422 0, 1423 /*40*/ MatAXPY_MPISBAIJ, 1424 MatGetSubMatrices_MPISBAIJ, 1425 MatIncreaseOverlap_MPISBAIJ, 1426 MatGetValues_MPISBAIJ, 1427 MatCopy_MPISBAIJ, 1428 /*45*/ 0, 1429 MatScale_MPISBAIJ, 1430 0, 1431 0, 1432 0, 1433 /*50*/ 0, 1434 0, 1435 0, 1436 0, 1437 0, 1438 /*55*/ 0, 1439 0, 1440 MatSetUnfactored_MPISBAIJ, 1441 0, 1442 MatSetValuesBlocked_MPISBAIJ, 1443 /*60*/ 0, 1444 0, 1445 0, 1446 0, 1447 0, 1448 /*65*/ 0, 1449 0, 1450 0, 1451 0, 1452 0, 1453 /*70*/ MatGetRowMaxAbs_MPISBAIJ, 1454 0, 1455 0, 1456 0, 1457 0, 1458 /*75*/ 0, 1459 0, 1460 0, 1461 0, 1462 0, 1463 /*80*/ 0, 1464 0, 1465 0, 1466 0, 1467 MatLoad_MPISBAIJ, 1468 /*85*/ 0, 1469 0, 1470 0, 1471 0, 1472 0, 1473 /*90*/ 0, 1474 0, 1475 0, 1476 0, 1477 0, 1478 /*95*/ 0, 1479 0, 1480 0, 1481 0, 1482 0, 1483 /*100*/0, 1484 0, 1485 0, 1486 0, 1487 0, 1488 /*105*/0, 1489 MatRealPart_MPISBAIJ, 1490 MatImaginaryPart_MPISBAIJ, 1491 MatGetRowUpperTriangular_MPISBAIJ, 1492 MatRestoreRowUpperTriangular_MPISBAIJ 1493 }; 1494 1495 1496 EXTERN_C_BEGIN 1497 #undef __FUNCT__ 1498 #define __FUNCT__ "MatGetDiagonalBlock_MPISBAIJ" 1499 PetscErrorCode PETSCMAT_DLLEXPORT MatGetDiagonalBlock_MPISBAIJ(Mat A,PetscTruth *iscopy,MatReuse reuse,Mat *a) 1500 { 1501 PetscFunctionBegin; 1502 *a = ((Mat_MPISBAIJ *)A->data)->A; 1503 *iscopy = PETSC_FALSE; 1504 PetscFunctionReturn(0); 1505 } 1506 EXTERN_C_END 1507 1508 EXTERN_C_BEGIN 1509 #undef __FUNCT__ 1510 #define __FUNCT__ "MatMPISBAIJSetPreallocation_MPISBAIJ" 1511 PetscErrorCode PETSCMAT_DLLEXPORT MatMPISBAIJSetPreallocation_MPISBAIJ(Mat B,PetscInt bs,PetscInt d_nz,PetscInt *d_nnz,PetscInt o_nz,PetscInt *o_nnz) 1512 { 1513 Mat_MPISBAIJ *b; 1514 PetscErrorCode ierr; 1515 PetscInt i,mbs,Mbs; 1516 1517 PetscFunctionBegin; 1518 ierr = PetscOptionsBegin(((PetscObject)B)->comm,((PetscObject)B)->prefix,"Options for MPISBAIJ matrix","Mat");CHKERRQ(ierr); 1519 ierr = PetscOptionsInt("-mat_block_size","Set the blocksize used to store the matrix","MatMPIBAIJSetPreallocation",bs,&bs,PETSC_NULL);CHKERRQ(ierr); 1520 ierr = PetscOptionsEnd();CHKERRQ(ierr); 1521 1522 if (bs < 1) SETERRQ(PETSC_ERR_ARG_OUTOFRANGE,"Invalid block size specified, must be positive"); 1523 if (d_nz == PETSC_DECIDE || d_nz == PETSC_DEFAULT) d_nz = 3; 1524 if (o_nz == PETSC_DECIDE || o_nz == PETSC_DEFAULT) o_nz = 1; 1525 if (d_nz < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"d_nz cannot be less than 0: value %D",d_nz); 1526 if (o_nz < 0) SETERRQ1(PETSC_ERR_ARG_OUTOFRANGE,"o_nz cannot be less than 0: value %D",o_nz); 1527 1528 B->rmap.bs = B->cmap.bs = bs; 1529 ierr = PetscMapSetUp(&B->rmap);CHKERRQ(ierr); 1530 ierr = PetscMapSetUp(&B->cmap);CHKERRQ(ierr); 1531 1532 if (d_nnz) { 1533 for (i=0; i<B->rmap.n/bs; i++) { 1534 if (d_nnz[i] < 0) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"d_nnz cannot be less than -1: local row %D value %D",i,d_nnz[i]); 1535 } 1536 } 1537 if (o_nnz) { 1538 for (i=0; i<B->rmap.n/bs; i++) { 1539 if (o_nnz[i] < 0) SETERRQ2(PETSC_ERR_ARG_OUTOFRANGE,"o_nnz cannot be less than -1: local row %D value %D",i,o_nnz[i]); 1540 } 1541 } 1542 B->preallocated = PETSC_TRUE; 1543 1544 b = (Mat_MPISBAIJ*)B->data; 1545 mbs = B->rmap.n/bs; 1546 Mbs = B->rmap.N/bs; 1547 if (mbs*bs != B->rmap.n) { 1548 SETERRQ2(PETSC_ERR_ARG_SIZ,"No of local rows %D must be divisible by blocksize %D",B->rmap.N,bs); 1549 } 1550 1551 B->rmap.bs = bs; 1552 b->bs2 = bs*bs; 1553 b->mbs = mbs; 1554 b->nbs = mbs; 1555 b->Mbs = Mbs; 1556 b->Nbs = Mbs; 1557 1558 for (i=0; i<=b->size; i++) { 1559 b->rangebs[i] = B->rmap.range[i]/bs; 1560 } 1561 b->rstartbs = B->rmap.rstart/bs; 1562 b->rendbs = B->rmap.rend/bs; 1563 1564 b->cstartbs = B->cmap.rstart/bs; 1565 b->cendbs = B->cmap.rend/bs; 1566 1567 ierr = MatCreate(PETSC_COMM_SELF,&b->A);CHKERRQ(ierr); 1568 ierr = MatSetSizes(b->A,B->rmap.n,B->cmap.n,B->rmap.n,B->cmap.n);CHKERRQ(ierr); 1569 ierr = MatSetType(b->A,MATSEQSBAIJ);CHKERRQ(ierr); 1570 ierr = MatSeqSBAIJSetPreallocation(b->A,bs,d_nz,d_nnz);CHKERRQ(ierr); 1571 ierr = PetscLogObjectParent(B,b->A);CHKERRQ(ierr); 1572 1573 ierr = MatCreate(PETSC_COMM_SELF,&b->B);CHKERRQ(ierr); 1574 ierr = MatSetSizes(b->B,B->rmap.n,B->cmap.N,B->rmap.n,B->cmap.N);CHKERRQ(ierr); 1575 ierr = MatSetType(b->B,MATSEQBAIJ);CHKERRQ(ierr); 1576 ierr = MatSeqBAIJSetPreallocation(b->B,bs,o_nz,o_nnz);CHKERRQ(ierr); 1577 ierr = PetscLogObjectParent(B,b->B);CHKERRQ(ierr); 1578 1579 /* build cache for off array entries formed */ 1580 ierr = MatStashCreate_Private(((PetscObject)B)->comm,bs,&B->bstash);CHKERRQ(ierr); 1581 1582 PetscFunctionReturn(0); 1583 } 1584 EXTERN_C_END 1585 1586 EXTERN_C_BEGIN 1587 extern PetscErrorCode MatGetFactor_mpisbaij_mumps(Mat,MatFactorType,Mat*); 1588 extern PetscErrorCode PETSCMAT_DLLEXPORT MatGetFactor_mpisbaij_spooles(Mat,MatFactorType,Mat*); 1589 EXTERN_C_END 1590 1591 /*MC 1592 MATMPISBAIJ - MATMPISBAIJ = "mpisbaij" - A matrix type to be used for distributed symmetric sparse block matrices, 1593 based on block compressed sparse row format. Only the upper triangular portion of the matrix is stored. 1594 1595 Options Database Keys: 1596 . -mat_type mpisbaij - sets the matrix type to "mpisbaij" during a call to MatSetFromOptions() 1597 1598 Level: beginner 1599 1600 .seealso: MatCreateMPISBAIJ 1601 M*/ 1602 1603 EXTERN_C_BEGIN 1604 #undef __FUNCT__ 1605 #define __FUNCT__ "MatCreate_MPISBAIJ" 1606 PetscErrorCode PETSCMAT_DLLEXPORT MatCreate_MPISBAIJ(Mat B) 1607 { 1608 Mat_MPISBAIJ *b; 1609 PetscErrorCode ierr; 1610 PetscTruth flg; 1611 1612 PetscFunctionBegin; 1613 1614 ierr = PetscNewLog(B,Mat_MPISBAIJ,&b);CHKERRQ(ierr); 1615 B->data = (void*)b; 1616 ierr = PetscMemcpy(B->ops,&MatOps_Values,sizeof(struct _MatOps));CHKERRQ(ierr); 1617 1618 B->ops->destroy = MatDestroy_MPISBAIJ; 1619 B->ops->view = MatView_MPISBAIJ; 1620 B->mapping = 0; 1621 B->assembled = PETSC_FALSE; 1622 1623 B->insertmode = NOT_SET_VALUES; 1624 ierr = MPI_Comm_rank(((PetscObject)B)->comm,&b->rank);CHKERRQ(ierr); 1625 ierr = MPI_Comm_size(((PetscObject)B)->comm,&b->size);CHKERRQ(ierr); 1626 1627 /* build local table of row and column ownerships */ 1628 ierr = PetscMalloc((b->size+2)*sizeof(PetscInt),&b->rangebs);CHKERRQ(ierr); 1629 1630 /* build cache for off array entries formed */ 1631 ierr = MatStashCreate_Private(((PetscObject)B)->comm,1,&B->stash);CHKERRQ(ierr); 1632 b->donotstash = PETSC_FALSE; 1633 b->colmap = PETSC_NULL; 1634 b->garray = PETSC_NULL; 1635 b->roworiented = PETSC_TRUE; 1636 1637 /* stuff used in block assembly */ 1638 b->barray = 0; 1639 1640 /* stuff used for matrix vector multiply */ 1641 b->lvec = 0; 1642 b->Mvctx = 0; 1643 b->slvec0 = 0; 1644 b->slvec0b = 0; 1645 b->slvec1 = 0; 1646 b->slvec1a = 0; 1647 b->slvec1b = 0; 1648 b->sMvctx = 0; 1649 1650 /* stuff for MatGetRow() */ 1651 b->rowindices = 0; 1652 b->rowvalues = 0; 1653 b->getrowactive = PETSC_FALSE; 1654 1655 /* hash table stuff */ 1656 b->ht = 0; 1657 b->hd = 0; 1658 b->ht_size = 0; 1659 b->ht_flag = PETSC_FALSE; 1660 b->ht_fact = 0; 1661 b->ht_total_ct = 0; 1662 b->ht_insert_ct = 0; 1663 1664 b->in_loc = 0; 1665 b->v_loc = 0; 1666 b->n_loc = 0; 1667 ierr = PetscOptionsBegin(((PetscObject)B)->comm,PETSC_NULL,"Options for loading MPISBAIJ matrix 1","Mat");CHKERRQ(ierr); 1668 ierr = PetscOptionsTruth("-mat_use_hash_table","Use hash table to save memory in constructing matrix","MatSetOption",PETSC_FALSE,&flg,PETSC_NULL);CHKERRQ(ierr); 1669 if (flg) { 1670 PetscReal fact = 1.39; 1671 ierr = MatSetOption(B,MAT_USE_HASH_TABLE,PETSC_TRUE);CHKERRQ(ierr); 1672 ierr = PetscOptionsReal("-mat_use_hash_table","Use hash table factor","MatMPIBAIJSetHashTableFactor",fact,&fact,PETSC_NULL);CHKERRQ(ierr); 1673 if (fact <= 1.0) fact = 1.39; 1674 ierr = MatMPIBAIJSetHashTableFactor(B,fact);CHKERRQ(ierr); 1675 ierr = PetscInfo1(B,"Hash table Factor used %5.2f\n",fact);CHKERRQ(ierr); 1676 } 1677 ierr = PetscOptionsEnd();CHKERRQ(ierr); 1678 1679 ierr = PetscObjectComposeFunctionDynamic((PetscObject)B,"MatGetFactor_mpisbaij_mumps_C", 1680 "MatGetFactor_mpisbaij_mumps", 1681 MatGetFactor_mpisbaij_mumps);CHKERRQ(ierr); 1682 ierr = PetscObjectComposeFunctionDynamic((PetscObject)B,"MatGetFactor_mpisbaij_spooles_C", 1683 "MatGetFactor_mpisbaij_spooles", 1684 MatGetFactor_mpisbaij_spooles);CHKERRQ(ierr); 1685 ierr = PetscObjectComposeFunctionDynamic((PetscObject)B,"MatStoreValues_C", 1686 "MatStoreValues_MPISBAIJ", 1687 MatStoreValues_MPISBAIJ);CHKERRQ(ierr); 1688 ierr = PetscObjectComposeFunctionDynamic((PetscObject)B,"MatRetrieveValues_C", 1689 "MatRetrieveValues_MPISBAIJ", 1690 MatRetrieveValues_MPISBAIJ);CHKERRQ(ierr); 1691 ierr = PetscObjectComposeFunctionDynamic((PetscObject)B,"MatGetDiagonalBlock_C", 1692 "MatGetDiagonalBlock_MPISBAIJ", 1693 MatGetDiagonalBlock_MPISBAIJ);CHKERRQ(ierr); 1694 ierr = PetscObjectComposeFunctionDynamic((PetscObject)B,"MatMPISBAIJSetPreallocation_C", 1695 "MatMPISBAIJSetPreallocation_MPISBAIJ", 1696 MatMPISBAIJSetPreallocation_MPISBAIJ);CHKERRQ(ierr); 1697 B->symmetric = PETSC_TRUE; 1698 B->structurally_symmetric = PETSC_TRUE; 1699 B->symmetric_set = PETSC_TRUE; 1700 B->structurally_symmetric_set = PETSC_TRUE; 1701 ierr = PetscObjectChangeTypeName((PetscObject)B,MATMPISBAIJ);CHKERRQ(ierr); 1702 PetscFunctionReturn(0); 1703 } 1704 EXTERN_C_END 1705 1706 /*MC 1707 MATSBAIJ - MATSBAIJ = "sbaij" - A matrix type to be used for symmetric block sparse matrices. 1708 1709 This matrix type is identical to MATSEQSBAIJ when constructed with a single process communicator, 1710 and MATMPISBAIJ otherwise. 1711 1712 Options Database Keys: 1713 . -mat_type sbaij - sets the matrix type to "sbaij" during a call to MatSetFromOptions() 1714 1715 Level: beginner 1716 1717 .seealso: MatCreateMPISBAIJ,MATSEQSBAIJ,MATMPISBAIJ 1718 M*/ 1719 1720 EXTERN_C_BEGIN 1721 #undef __FUNCT__ 1722 #define __FUNCT__ "MatCreate_SBAIJ" 1723 PetscErrorCode PETSCMAT_DLLEXPORT MatCreate_SBAIJ(Mat A) 1724 { 1725 PetscErrorCode ierr; 1726 PetscMPIInt size; 1727 1728 PetscFunctionBegin; 1729 ierr = MPI_Comm_size(((PetscObject)A)->comm,&size);CHKERRQ(ierr); 1730 if (size == 1) { 1731 ierr = MatSetType(A,MATSEQSBAIJ);CHKERRQ(ierr); 1732 } else { 1733 ierr = MatSetType(A,MATMPISBAIJ);CHKERRQ(ierr); 1734 } 1735 PetscFunctionReturn(0); 1736 } 1737 EXTERN_C_END 1738 1739 #undef __FUNCT__ 1740 #define __FUNCT__ "MatMPISBAIJSetPreallocation" 1741 /*@C 1742 MatMPISBAIJSetPreallocation - For good matrix assembly performance 1743 the user should preallocate the matrix storage by setting the parameters 1744 d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately, 1745 performance can be increased by more than a factor of 50. 1746 1747 Collective on Mat 1748 1749 Input Parameters: 1750 + A - the matrix 1751 . bs - size of blockk 1752 . d_nz - number of block nonzeros per block row in diagonal portion of local 1753 submatrix (same for all local rows) 1754 . d_nnz - array containing the number of block nonzeros in the various block rows 1755 in the upper triangular and diagonal part of the in diagonal portion of the local 1756 (possibly different for each block row) or PETSC_NULL. You must leave room 1757 for the diagonal entry even if it is zero. 1758 . o_nz - number of block nonzeros per block row in the off-diagonal portion of local 1759 submatrix (same for all local rows). 1760 - o_nnz - array containing the number of nonzeros in the various block rows of the 1761 off-diagonal portion of the local submatrix (possibly different for 1762 each block row) or PETSC_NULL. 1763 1764 1765 Options Database Keys: 1766 . -mat_no_unroll - uses code that does not unroll the loops in the 1767 block calculations (much slower) 1768 . -mat_block_size - size of the blocks to use 1769 1770 Notes: 1771 1772 If PETSC_DECIDE or PETSC_DETERMINE is used for a particular argument on one processor 1773 than it must be used on all processors that share the object for that argument. 1774 1775 If the *_nnz parameter is given then the *_nz parameter is ignored 1776 1777 Storage Information: 1778 For a square global matrix we define each processor's diagonal portion 1779 to be its local rows and the corresponding columns (a square submatrix); 1780 each processor's off-diagonal portion encompasses the remainder of the 1781 local matrix (a rectangular submatrix). 1782 1783 The user can specify preallocated storage for the diagonal part of 1784 the local submatrix with either d_nz or d_nnz (not both). Set 1785 d_nz=PETSC_DEFAULT and d_nnz=PETSC_NULL for PETSc to control dynamic 1786 memory allocation. Likewise, specify preallocated storage for the 1787 off-diagonal part of the local submatrix with o_nz or o_nnz (not both). 1788 1789 You can call MatGetInfo() to get information on how effective the preallocation was; 1790 for example the fields mallocs,nz_allocated,nz_used,nz_unneeded; 1791 You can also run with the option -info and look for messages with the string 1792 malloc in them to see if additional memory allocation was needed. 1793 1794 Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In 1795 the figure below we depict these three local rows and all columns (0-11). 1796 1797 .vb 1798 0 1 2 3 4 5 6 7 8 9 10 11 1799 ------------------- 1800 row 3 | o o o d d d o o o o o o 1801 row 4 | o o o d d d o o o o o o 1802 row 5 | o o o d d d o o o o o o 1803 ------------------- 1804 .ve 1805 1806 Thus, any entries in the d locations are stored in the d (diagonal) 1807 submatrix, and any entries in the o locations are stored in the 1808 o (off-diagonal) submatrix. Note that the d matrix is stored in 1809 MatSeqSBAIJ format and the o submatrix in MATSEQBAIJ format. 1810 1811 Now d_nz should indicate the number of block nonzeros per row in the upper triangular 1812 plus the diagonal part of the d matrix, 1813 and o_nz should indicate the number of block nonzeros per row in the o matrix. 1814 In general, for PDE problems in which most nonzeros are near the diagonal, 1815 one expects d_nz >> o_nz. For large problems you MUST preallocate memory 1816 or you will get TERRIBLE performance; see the users' manual chapter on 1817 matrices. 1818 1819 Level: intermediate 1820 1821 .keywords: matrix, block, aij, compressed row, sparse, parallel 1822 1823 .seealso: MatCreate(), MatCreateSeqSBAIJ(), MatSetValues(), MatCreateMPIBAIJ() 1824 @*/ 1825 PetscErrorCode PETSCMAT_DLLEXPORT MatMPISBAIJSetPreallocation(Mat B,PetscInt bs,PetscInt d_nz,const PetscInt d_nnz[],PetscInt o_nz,const PetscInt o_nnz[]) 1826 { 1827 PetscErrorCode ierr,(*f)(Mat,PetscInt,PetscInt,const PetscInt[],PetscInt,const PetscInt[]); 1828 1829 PetscFunctionBegin; 1830 ierr = PetscObjectQueryFunction((PetscObject)B,"MatMPISBAIJSetPreallocation_C",(void (**)(void))&f);CHKERRQ(ierr); 1831 if (f) { 1832 ierr = (*f)(B,bs,d_nz,d_nnz,o_nz,o_nnz);CHKERRQ(ierr); 1833 } 1834 PetscFunctionReturn(0); 1835 } 1836 1837 #undef __FUNCT__ 1838 #define __FUNCT__ "MatCreateMPISBAIJ" 1839 /*@C 1840 MatCreateMPISBAIJ - Creates a sparse parallel matrix in symmetric block AIJ format 1841 (block compressed row). For good matrix assembly performance 1842 the user should preallocate the matrix storage by setting the parameters 1843 d_nz (or d_nnz) and o_nz (or o_nnz). By setting these parameters accurately, 1844 performance can be increased by more than a factor of 50. 1845 1846 Collective on MPI_Comm 1847 1848 Input Parameters: 1849 + comm - MPI communicator 1850 . bs - size of blockk 1851 . m - number of local rows (or PETSC_DECIDE to have calculated if M is given) 1852 This value should be the same as the local size used in creating the 1853 y vector for the matrix-vector product y = Ax. 1854 . n - number of local columns (or PETSC_DECIDE to have calculated if N is given) 1855 This value should be the same as the local size used in creating the 1856 x vector for the matrix-vector product y = Ax. 1857 . M - number of global rows (or PETSC_DETERMINE to have calculated if m is given) 1858 . N - number of global columns (or PETSC_DETERMINE to have calculated if n is given) 1859 . d_nz - number of block nonzeros per block row in diagonal portion of local 1860 submatrix (same for all local rows) 1861 . d_nnz - array containing the number of block nonzeros in the various block rows 1862 in the upper triangular portion of the in diagonal portion of the local 1863 (possibly different for each block block row) or PETSC_NULL. 1864 You must leave room for the diagonal entry even if it is zero. 1865 . o_nz - number of block nonzeros per block row in the off-diagonal portion of local 1866 submatrix (same for all local rows). 1867 - o_nnz - array containing the number of nonzeros in the various block rows of the 1868 off-diagonal portion of the local submatrix (possibly different for 1869 each block row) or PETSC_NULL. 1870 1871 Output Parameter: 1872 . A - the matrix 1873 1874 Options Database Keys: 1875 . -mat_no_unroll - uses code that does not unroll the loops in the 1876 block calculations (much slower) 1877 . -mat_block_size - size of the blocks to use 1878 . -mat_mpi - use the parallel matrix data structures even on one processor 1879 (defaults to using SeqBAIJ format on one processor) 1880 1881 It is recommended that one use the MatCreate(), MatSetType() and/or MatSetFromOptions(), 1882 MatXXXXSetPreallocation() paradgm instead of this routine directly. This is definitely 1883 true if you plan to use the external direct solvers such as SuperLU, MUMPS or Spooles. 1884 [MatXXXXSetPreallocation() is, for example, MatSeqAIJSetPreallocation] 1885 1886 Notes: 1887 The number of rows and columns must be divisible by blocksize. 1888 This matrix type does not support complex Hermitian operation. 1889 1890 The user MUST specify either the local or global matrix dimensions 1891 (possibly both). 1892 1893 If PETSC_DECIDE or PETSC_DETERMINE is used for a particular argument on one processor 1894 than it must be used on all processors that share the object for that argument. 1895 1896 If the *_nnz parameter is given then the *_nz parameter is ignored 1897 1898 Storage Information: 1899 For a square global matrix we define each processor's diagonal portion 1900 to be its local rows and the corresponding columns (a square submatrix); 1901 each processor's off-diagonal portion encompasses the remainder of the 1902 local matrix (a rectangular submatrix). 1903 1904 The user can specify preallocated storage for the diagonal part of 1905 the local submatrix with either d_nz or d_nnz (not both). Set 1906 d_nz=PETSC_DEFAULT and d_nnz=PETSC_NULL for PETSc to control dynamic 1907 memory allocation. Likewise, specify preallocated storage for the 1908 off-diagonal part of the local submatrix with o_nz or o_nnz (not both). 1909 1910 Consider a processor that owns rows 3, 4 and 5 of a parallel matrix. In 1911 the figure below we depict these three local rows and all columns (0-11). 1912 1913 .vb 1914 0 1 2 3 4 5 6 7 8 9 10 11 1915 ------------------- 1916 row 3 | o o o d d d o o o o o o 1917 row 4 | o o o d d d o o o o o o 1918 row 5 | o o o d d d o o o o o o 1919 ------------------- 1920 .ve 1921 1922 Thus, any entries in the d locations are stored in the d (diagonal) 1923 submatrix, and any entries in the o locations are stored in the 1924 o (off-diagonal) submatrix. Note that the d matrix is stored in 1925 MatSeqSBAIJ format and the o submatrix in MATSEQBAIJ format. 1926 1927 Now d_nz should indicate the number of block nonzeros per row in the upper triangular 1928 plus the diagonal part of the d matrix, 1929 and o_nz should indicate the number of block nonzeros per row in the o matrix. 1930 In general, for PDE problems in which most nonzeros are near the diagonal, 1931 one expects d_nz >> o_nz. For large problems you MUST preallocate memory 1932 or you will get TERRIBLE performance; see the users' manual chapter on 1933 matrices. 1934 1935 Level: intermediate 1936 1937 .keywords: matrix, block, aij, compressed row, sparse, parallel 1938 1939 .seealso: MatCreate(), MatCreateSeqSBAIJ(), MatSetValues(), MatCreateMPIBAIJ() 1940 @*/ 1941 1942 PetscErrorCode PETSCMAT_DLLEXPORT MatCreateMPISBAIJ(MPI_Comm comm,PetscInt bs,PetscInt m,PetscInt n,PetscInt M,PetscInt N,PetscInt d_nz,const PetscInt d_nnz[],PetscInt o_nz,const PetscInt o_nnz[],Mat *A) 1943 { 1944 PetscErrorCode ierr; 1945 PetscMPIInt size; 1946 1947 PetscFunctionBegin; 1948 ierr = MatCreate(comm,A);CHKERRQ(ierr); 1949 ierr = MatSetSizes(*A,m,n,M,N);CHKERRQ(ierr); 1950 ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr); 1951 if (size > 1) { 1952 ierr = MatSetType(*A,MATMPISBAIJ);CHKERRQ(ierr); 1953 ierr = MatMPISBAIJSetPreallocation(*A,bs,d_nz,d_nnz,o_nz,o_nnz);CHKERRQ(ierr); 1954 } else { 1955 ierr = MatSetType(*A,MATSEQSBAIJ);CHKERRQ(ierr); 1956 ierr = MatSeqSBAIJSetPreallocation(*A,bs,d_nz,d_nnz);CHKERRQ(ierr); 1957 } 1958 PetscFunctionReturn(0); 1959 } 1960 1961 1962 #undef __FUNCT__ 1963 #define __FUNCT__ "MatDuplicate_MPISBAIJ" 1964 static PetscErrorCode MatDuplicate_MPISBAIJ(Mat matin,MatDuplicateOption cpvalues,Mat *newmat) 1965 { 1966 Mat mat; 1967 Mat_MPISBAIJ *a,*oldmat = (Mat_MPISBAIJ*)matin->data; 1968 PetscErrorCode ierr; 1969 PetscInt len=0,nt,bs=matin->rmap.bs,mbs=oldmat->mbs; 1970 PetscScalar *array; 1971 1972 PetscFunctionBegin; 1973 *newmat = 0; 1974 ierr = MatCreate(((PetscObject)matin)->comm,&mat);CHKERRQ(ierr); 1975 ierr = MatSetSizes(mat,matin->rmap.n,matin->cmap.n,matin->rmap.N,matin->cmap.N);CHKERRQ(ierr); 1976 ierr = MatSetType(mat,((PetscObject)matin)->type_name);CHKERRQ(ierr); 1977 ierr = PetscMemcpy(mat->ops,matin->ops,sizeof(struct _MatOps));CHKERRQ(ierr); 1978 ierr = PetscMapCopy(((PetscObject)matin)->comm,&matin->rmap,&mat->rmap);CHKERRQ(ierr); 1979 ierr = PetscMapCopy(((PetscObject)matin)->comm,&matin->cmap,&mat->cmap);CHKERRQ(ierr); 1980 1981 mat->factor = matin->factor; 1982 mat->preallocated = PETSC_TRUE; 1983 mat->assembled = PETSC_TRUE; 1984 mat->insertmode = NOT_SET_VALUES; 1985 1986 a = (Mat_MPISBAIJ*)mat->data; 1987 a->bs2 = oldmat->bs2; 1988 a->mbs = oldmat->mbs; 1989 a->nbs = oldmat->nbs; 1990 a->Mbs = oldmat->Mbs; 1991 a->Nbs = oldmat->Nbs; 1992 1993 1994 a->size = oldmat->size; 1995 a->rank = oldmat->rank; 1996 a->donotstash = oldmat->donotstash; 1997 a->roworiented = oldmat->roworiented; 1998 a->rowindices = 0; 1999 a->rowvalues = 0; 2000 a->getrowactive = PETSC_FALSE; 2001 a->barray = 0; 2002 a->rstartbs = oldmat->rstartbs; 2003 a->rendbs = oldmat->rendbs; 2004 a->cstartbs = oldmat->cstartbs; 2005 a->cendbs = oldmat->cendbs; 2006 2007 /* hash table stuff */ 2008 a->ht = 0; 2009 a->hd = 0; 2010 a->ht_size = 0; 2011 a->ht_flag = oldmat->ht_flag; 2012 a->ht_fact = oldmat->ht_fact; 2013 a->ht_total_ct = 0; 2014 a->ht_insert_ct = 0; 2015 2016 ierr = PetscMemcpy(a->rangebs,oldmat->rangebs,(a->size+2)*sizeof(PetscInt));CHKERRQ(ierr); 2017 ierr = MatStashCreate_Private(((PetscObject)matin)->comm,1,&mat->stash);CHKERRQ(ierr); 2018 ierr = MatStashCreate_Private(((PetscObject)matin)->comm,matin->rmap.bs,&mat->bstash);CHKERRQ(ierr); 2019 if (oldmat->colmap) { 2020 #if defined (PETSC_USE_CTABLE) 2021 ierr = PetscTableCreateCopy(oldmat->colmap,&a->colmap);CHKERRQ(ierr); 2022 #else 2023 ierr = PetscMalloc((a->Nbs)*sizeof(PetscInt),&a->colmap);CHKERRQ(ierr); 2024 ierr = PetscLogObjectMemory(mat,(a->Nbs)*sizeof(PetscInt));CHKERRQ(ierr); 2025 ierr = PetscMemcpy(a->colmap,oldmat->colmap,(a->Nbs)*sizeof(PetscInt));CHKERRQ(ierr); 2026 #endif 2027 } else a->colmap = 0; 2028 2029 if (oldmat->garray && (len = ((Mat_SeqBAIJ*)(oldmat->B->data))->nbs)) { 2030 ierr = PetscMalloc(len*sizeof(PetscInt),&a->garray);CHKERRQ(ierr); 2031 ierr = PetscLogObjectMemory(mat,len*sizeof(PetscInt));CHKERRQ(ierr); 2032 ierr = PetscMemcpy(a->garray,oldmat->garray,len*sizeof(PetscInt));CHKERRQ(ierr); 2033 } else a->garray = 0; 2034 2035 ierr = VecDuplicate(oldmat->lvec,&a->lvec);CHKERRQ(ierr); 2036 ierr = PetscLogObjectParent(mat,a->lvec);CHKERRQ(ierr); 2037 ierr = VecScatterCopy(oldmat->Mvctx,&a->Mvctx);CHKERRQ(ierr); 2038 ierr = PetscLogObjectParent(mat,a->Mvctx);CHKERRQ(ierr); 2039 2040 ierr = VecDuplicate(oldmat->slvec0,&a->slvec0);CHKERRQ(ierr); 2041 ierr = PetscLogObjectParent(mat,a->slvec0);CHKERRQ(ierr); 2042 ierr = VecDuplicate(oldmat->slvec1,&a->slvec1);CHKERRQ(ierr); 2043 ierr = PetscLogObjectParent(mat,a->slvec1);CHKERRQ(ierr); 2044 2045 ierr = VecGetLocalSize(a->slvec1,&nt);CHKERRQ(ierr); 2046 ierr = VecGetArray(a->slvec1,&array);CHKERRQ(ierr); 2047 ierr = VecCreateSeqWithArray(PETSC_COMM_SELF,bs*mbs,array,&a->slvec1a);CHKERRQ(ierr); 2048 ierr = VecCreateSeqWithArray(PETSC_COMM_SELF,nt-bs*mbs,array+bs*mbs,&a->slvec1b);CHKERRQ(ierr); 2049 ierr = VecRestoreArray(a->slvec1,&array);CHKERRQ(ierr); 2050 ierr = VecGetArray(a->slvec0,&array);CHKERRQ(ierr); 2051 ierr = VecCreateSeqWithArray(PETSC_COMM_SELF,nt-bs*mbs,array+bs*mbs,&a->slvec0b);CHKERRQ(ierr); 2052 ierr = VecRestoreArray(a->slvec0,&array);CHKERRQ(ierr); 2053 ierr = PetscLogObjectParent(mat,a->slvec0);CHKERRQ(ierr); 2054 ierr = PetscLogObjectParent(mat,a->slvec1);CHKERRQ(ierr); 2055 ierr = PetscLogObjectParent(mat,a->slvec0b);CHKERRQ(ierr); 2056 ierr = PetscLogObjectParent(mat,a->slvec1a);CHKERRQ(ierr); 2057 ierr = PetscLogObjectParent(mat,a->slvec1b);CHKERRQ(ierr); 2058 2059 /* ierr = VecScatterCopy(oldmat->sMvctx,&a->sMvctx); - not written yet, replaced by the lazy trick: */ 2060 ierr = PetscObjectReference((PetscObject)oldmat->sMvctx);CHKERRQ(ierr); 2061 a->sMvctx = oldmat->sMvctx; 2062 ierr = PetscLogObjectParent(mat,a->sMvctx);CHKERRQ(ierr); 2063 2064 ierr = MatDuplicate(oldmat->A,cpvalues,&a->A);CHKERRQ(ierr); 2065 ierr = PetscLogObjectParent(mat,a->A);CHKERRQ(ierr); 2066 ierr = MatDuplicate(oldmat->B,cpvalues,&a->B);CHKERRQ(ierr); 2067 ierr = PetscLogObjectParent(mat,a->B);CHKERRQ(ierr); 2068 ierr = PetscFListDuplicate(((PetscObject)matin)->qlist,&((PetscObject)mat)->qlist);CHKERRQ(ierr); 2069 *newmat = mat; 2070 PetscFunctionReturn(0); 2071 } 2072 2073 #include "petscsys.h" 2074 2075 #undef __FUNCT__ 2076 #define __FUNCT__ "MatLoad_MPISBAIJ" 2077 PetscErrorCode MatLoad_MPISBAIJ(PetscViewer viewer, MatType type,Mat *newmat) 2078 { 2079 Mat A; 2080 PetscErrorCode ierr; 2081 PetscInt i,nz,j,rstart,rend; 2082 PetscScalar *vals,*buf; 2083 MPI_Comm comm = ((PetscObject)viewer)->comm; 2084 MPI_Status status; 2085 PetscMPIInt rank,size,tag = ((PetscObject)viewer)->tag,*sndcounts = 0,*browners,maxnz,*rowners,*locrowlens,mmbs; 2086 PetscInt header[4],*rowlengths = 0,M,N,m,*cols; 2087 PetscInt *procsnz = 0,jj,*mycols,*ibuf; 2088 PetscInt bs=1,Mbs,mbs,extra_rows; 2089 PetscInt *dlens,*odlens,*mask,*masked1,*masked2,rowcount,odcount; 2090 PetscInt dcount,kmax,k,nzcount,tmp; 2091 int fd; 2092 2093 PetscFunctionBegin; 2094 ierr = PetscOptionsBegin(comm,PETSC_NULL,"Options for loading MPISBAIJ matrix 2","Mat");CHKERRQ(ierr); 2095 ierr = PetscOptionsInt("-matload_block_size","Set the blocksize used to store the matrix","MatLoad",bs,&bs,PETSC_NULL);CHKERRQ(ierr); 2096 ierr = PetscOptionsEnd();CHKERRQ(ierr); 2097 2098 ierr = MPI_Comm_size(comm,&size);CHKERRQ(ierr); 2099 ierr = MPI_Comm_rank(comm,&rank);CHKERRQ(ierr); 2100 if (!rank) { 2101 ierr = PetscViewerBinaryGetDescriptor(viewer,&fd);CHKERRQ(ierr); 2102 ierr = PetscBinaryRead(fd,(char *)header,4,PETSC_INT);CHKERRQ(ierr); 2103 if (header[0] != MAT_FILE_COOKIE) SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"not matrix object"); 2104 if (header[3] < 0) { 2105 SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"Matrix stored in special format, cannot load as MPISBAIJ"); 2106 } 2107 } 2108 2109 ierr = MPI_Bcast(header+1,3,MPIU_INT,0,comm);CHKERRQ(ierr); 2110 M = header[1]; N = header[2]; 2111 2112 if (M != N) SETERRQ(PETSC_ERR_SUP,"Can only do square matrices"); 2113 2114 /* 2115 This code adds extra rows to make sure the number of rows is 2116 divisible by the blocksize 2117 */ 2118 Mbs = M/bs; 2119 extra_rows = bs - M + bs*(Mbs); 2120 if (extra_rows == bs) extra_rows = 0; 2121 else Mbs++; 2122 if (extra_rows &&!rank) { 2123 ierr = PetscInfo(viewer,"Padding loaded matrix to match blocksize\n");CHKERRQ(ierr); 2124 } 2125 2126 /* determine ownership of all rows */ 2127 mbs = Mbs/size + ((Mbs % size) > rank); 2128 m = mbs*bs; 2129 ierr = PetscMalloc(2*(size+2)*sizeof(PetscMPIInt),&rowners);CHKERRQ(ierr); 2130 browners = rowners + size + 1; 2131 mmbs = PetscMPIIntCast(mbs); 2132 ierr = MPI_Allgather(&mmbs,1,MPI_INT,rowners+1,1,MPI_INT,comm);CHKERRQ(ierr); 2133 rowners[0] = 0; 2134 for (i=2; i<=size; i++) rowners[i] += rowners[i-1]; 2135 for (i=0; i<=size; i++) browners[i] = rowners[i]*bs; 2136 rstart = rowners[rank]; 2137 rend = rowners[rank+1]; 2138 2139 /* distribute row lengths to all processors */ 2140 ierr = PetscMalloc((rend-rstart)*bs*sizeof(PetscMPIInt),&locrowlens);CHKERRQ(ierr); 2141 if (!rank) { 2142 ierr = PetscMalloc((M+extra_rows)*sizeof(PetscInt),&rowlengths);CHKERRQ(ierr); 2143 ierr = PetscBinaryRead(fd,rowlengths,M,PETSC_INT);CHKERRQ(ierr); 2144 for (i=0; i<extra_rows; i++) rowlengths[M+i] = 1; 2145 ierr = PetscMalloc(size*sizeof(PetscMPIInt),&sndcounts);CHKERRQ(ierr); 2146 for (i=0; i<size; i++) sndcounts[i] = browners[i+1] - browners[i]; 2147 ierr = MPI_Scatterv(rowlengths,sndcounts,browners,MPIU_INT,locrowlens,(rend-rstart)*bs,MPIU_INT,0,comm);CHKERRQ(ierr); 2148 ierr = PetscFree(sndcounts);CHKERRQ(ierr); 2149 } else { 2150 ierr = MPI_Scatterv(0,0,0,MPIU_INT,locrowlens,(rend-rstart)*bs,MPIU_INT,0,comm);CHKERRQ(ierr); 2151 } 2152 2153 if (!rank) { /* procs[0] */ 2154 /* calculate the number of nonzeros on each processor */ 2155 ierr = PetscMalloc(size*sizeof(PetscInt),&procsnz);CHKERRQ(ierr); 2156 ierr = PetscMemzero(procsnz,size*sizeof(PetscInt));CHKERRQ(ierr); 2157 for (i=0; i<size; i++) { 2158 for (j=rowners[i]*bs; j< rowners[i+1]*bs; j++) { 2159 procsnz[i] += rowlengths[j]; 2160 } 2161 } 2162 ierr = PetscFree(rowlengths);CHKERRQ(ierr); 2163 2164 /* determine max buffer needed and allocate it */ 2165 maxnz = 0; 2166 for (i=0; i<size; i++) { 2167 maxnz = PetscMax(maxnz,procsnz[i]); 2168 } 2169 ierr = PetscMalloc(maxnz*sizeof(PetscInt),&cols);CHKERRQ(ierr); 2170 2171 /* read in my part of the matrix column indices */ 2172 nz = procsnz[0]; 2173 ierr = PetscMalloc(nz*sizeof(PetscInt),&ibuf);CHKERRQ(ierr); 2174 mycols = ibuf; 2175 if (size == 1) nz -= extra_rows; 2176 ierr = PetscBinaryRead(fd,mycols,nz,PETSC_INT);CHKERRQ(ierr); 2177 if (size == 1) for (i=0; i< extra_rows; i++) { mycols[nz+i] = M+i; } 2178 2179 /* read in every ones (except the last) and ship off */ 2180 for (i=1; i<size-1; i++) { 2181 nz = procsnz[i]; 2182 ierr = PetscBinaryRead(fd,cols,nz,PETSC_INT);CHKERRQ(ierr); 2183 ierr = MPI_Send(cols,nz,MPIU_INT,i,tag,comm);CHKERRQ(ierr); 2184 } 2185 /* read in the stuff for the last proc */ 2186 if (size != 1) { 2187 nz = procsnz[size-1] - extra_rows; /* the extra rows are not on the disk */ 2188 ierr = PetscBinaryRead(fd,cols,nz,PETSC_INT);CHKERRQ(ierr); 2189 for (i=0; i<extra_rows; i++) cols[nz+i] = M+i; 2190 ierr = MPI_Send(cols,nz+extra_rows,MPIU_INT,size-1,tag,comm);CHKERRQ(ierr); 2191 } 2192 ierr = PetscFree(cols);CHKERRQ(ierr); 2193 } else { /* procs[i], i>0 */ 2194 /* determine buffer space needed for message */ 2195 nz = 0; 2196 for (i=0; i<m; i++) { 2197 nz += locrowlens[i]; 2198 } 2199 ierr = PetscMalloc(nz*sizeof(PetscInt),&ibuf);CHKERRQ(ierr); 2200 mycols = ibuf; 2201 /* receive message of column indices*/ 2202 ierr = MPI_Recv(mycols,nz,MPIU_INT,0,tag,comm,&status);CHKERRQ(ierr); 2203 ierr = MPI_Get_count(&status,MPIU_INT,&maxnz);CHKERRQ(ierr); 2204 if (maxnz != nz) SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"something is wrong with file"); 2205 } 2206 2207 /* loop over local rows, determining number of off diagonal entries */ 2208 ierr = PetscMalloc(2*(rend-rstart+1)*sizeof(PetscInt),&dlens);CHKERRQ(ierr); 2209 odlens = dlens + (rend-rstart); 2210 ierr = PetscMalloc(3*Mbs*sizeof(PetscInt),&mask);CHKERRQ(ierr); 2211 ierr = PetscMemzero(mask,3*Mbs*sizeof(PetscInt));CHKERRQ(ierr); 2212 masked1 = mask + Mbs; 2213 masked2 = masked1 + Mbs; 2214 rowcount = 0; nzcount = 0; 2215 for (i=0; i<mbs; i++) { 2216 dcount = 0; 2217 odcount = 0; 2218 for (j=0; j<bs; j++) { 2219 kmax = locrowlens[rowcount]; 2220 for (k=0; k<kmax; k++) { 2221 tmp = mycols[nzcount++]/bs; /* block col. index */ 2222 if (!mask[tmp]) { 2223 mask[tmp] = 1; 2224 if (tmp < rstart || tmp >= rend) masked2[odcount++] = tmp; /* entry in off-diag portion */ 2225 else masked1[dcount++] = tmp; /* entry in diag portion */ 2226 } 2227 } 2228 rowcount++; 2229 } 2230 2231 dlens[i] = dcount; /* d_nzz[i] */ 2232 odlens[i] = odcount; /* o_nzz[i] */ 2233 2234 /* zero out the mask elements we set */ 2235 for (j=0; j<dcount; j++) mask[masked1[j]] = 0; 2236 for (j=0; j<odcount; j++) mask[masked2[j]] = 0; 2237 } 2238 2239 /* create our matrix */ 2240 ierr = MatCreate(comm,&A);CHKERRQ(ierr); 2241 ierr = MatSetSizes(A,m,m,PETSC_DETERMINE,PETSC_DETERMINE);CHKERRQ(ierr); 2242 ierr = MatSetType(A,type);CHKERRQ(ierr); 2243 ierr = MatMPISBAIJSetPreallocation(A,bs,0,dlens,0,odlens);CHKERRQ(ierr); 2244 2245 if (!rank) { 2246 ierr = PetscMalloc(maxnz*sizeof(PetscScalar),&buf);CHKERRQ(ierr); 2247 /* read in my part of the matrix numerical values */ 2248 nz = procsnz[0]; 2249 vals = buf; 2250 mycols = ibuf; 2251 if (size == 1) nz -= extra_rows; 2252 ierr = PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);CHKERRQ(ierr); 2253 if (size == 1) for (i=0; i< extra_rows; i++) { vals[nz+i] = 1.0; } 2254 2255 /* insert into matrix */ 2256 jj = rstart*bs; 2257 for (i=0; i<m; i++) { 2258 ierr = MatSetValues(A,1,&jj,locrowlens[i],mycols,vals,INSERT_VALUES);CHKERRQ(ierr); 2259 mycols += locrowlens[i]; 2260 vals += locrowlens[i]; 2261 jj++; 2262 } 2263 2264 /* read in other processors (except the last one) and ship out */ 2265 for (i=1; i<size-1; i++) { 2266 nz = procsnz[i]; 2267 vals = buf; 2268 ierr = PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);CHKERRQ(ierr); 2269 ierr = MPI_Send(vals,nz,MPIU_SCALAR,i,((PetscObject)A)->tag,comm);CHKERRQ(ierr); 2270 } 2271 /* the last proc */ 2272 if (size != 1){ 2273 nz = procsnz[i] - extra_rows; 2274 vals = buf; 2275 ierr = PetscBinaryRead(fd,vals,nz,PETSC_SCALAR);CHKERRQ(ierr); 2276 for (i=0; i<extra_rows; i++) vals[nz+i] = 1.0; 2277 ierr = MPI_Send(vals,nz+extra_rows,MPIU_SCALAR,size-1,((PetscObject)A)->tag,comm);CHKERRQ(ierr); 2278 } 2279 ierr = PetscFree(procsnz);CHKERRQ(ierr); 2280 2281 } else { 2282 /* receive numeric values */ 2283 ierr = PetscMalloc(nz*sizeof(PetscScalar),&buf);CHKERRQ(ierr); 2284 2285 /* receive message of values*/ 2286 vals = buf; 2287 mycols = ibuf; 2288 ierr = MPI_Recv(vals,nz,MPIU_SCALAR,0,((PetscObject)A)->tag,comm,&status);CHKERRQ(ierr); 2289 ierr = MPI_Get_count(&status,MPIU_SCALAR,&maxnz);CHKERRQ(ierr); 2290 if (maxnz != nz) SETERRQ(PETSC_ERR_FILE_UNEXPECTED,"something is wrong with file"); 2291 2292 /* insert into matrix */ 2293 jj = rstart*bs; 2294 for (i=0; i<m; i++) { 2295 ierr = MatSetValues_MPISBAIJ(A,1,&jj,locrowlens[i],mycols,vals,INSERT_VALUES);CHKERRQ(ierr); 2296 mycols += locrowlens[i]; 2297 vals += locrowlens[i]; 2298 jj++; 2299 } 2300 } 2301 2302 ierr = PetscFree(locrowlens);CHKERRQ(ierr); 2303 ierr = PetscFree(buf);CHKERRQ(ierr); 2304 ierr = PetscFree(ibuf);CHKERRQ(ierr); 2305 ierr = PetscFree(rowners);CHKERRQ(ierr); 2306 ierr = PetscFree(dlens);CHKERRQ(ierr); 2307 ierr = PetscFree(mask);CHKERRQ(ierr); 2308 ierr = MatAssemblyBegin(A,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2309 ierr = MatAssemblyEnd(A,MAT_FINAL_ASSEMBLY);CHKERRQ(ierr); 2310 *newmat = A; 2311 PetscFunctionReturn(0); 2312 } 2313 2314 #undef __FUNCT__ 2315 #define __FUNCT__ "MatMPISBAIJSetHashTableFactor" 2316 /*XXXXX@ 2317 MatMPISBAIJSetHashTableFactor - Sets the factor required to compute the size of the HashTable. 2318 2319 Input Parameters: 2320 . mat - the matrix 2321 . fact - factor 2322 2323 Collective on Mat 2324 2325 Level: advanced 2326 2327 Notes: 2328 This can also be set by the command line option: -mat_use_hash_table fact 2329 2330 .keywords: matrix, hashtable, factor, HT 2331 2332 .seealso: MatSetOption() 2333 @XXXXX*/ 2334 2335 2336 #undef __FUNCT__ 2337 #define __FUNCT__ "MatGetRowMaxAbs_MPISBAIJ" 2338 PetscErrorCode MatGetRowMaxAbs_MPISBAIJ(Mat A,Vec v,PetscInt idx[]) 2339 { 2340 Mat_MPISBAIJ *a = (Mat_MPISBAIJ*)A->data; 2341 Mat_SeqBAIJ *b = (Mat_SeqBAIJ*)(a->B)->data; 2342 PetscReal atmp; 2343 PetscReal *work,*svalues,*rvalues; 2344 PetscErrorCode ierr; 2345 PetscInt i,bs,mbs,*bi,*bj,brow,j,ncols,krow,kcol,col,row,Mbs,bcol; 2346 PetscMPIInt rank,size; 2347 PetscInt *rowners_bs,dest,count,source; 2348 PetscScalar *va; 2349 MatScalar *ba; 2350 MPI_Status stat; 2351 2352 PetscFunctionBegin; 2353 if (idx) SETERRQ(PETSC_ERR_SUP,"Send email to petsc-maint@mcs.anl.gov"); 2354 ierr = MatGetRowMaxAbs(a->A,v,PETSC_NULL);CHKERRQ(ierr); 2355 ierr = VecGetArray(v,&va);CHKERRQ(ierr); 2356 2357 ierr = MPI_Comm_size(((PetscObject)A)->comm,&size);CHKERRQ(ierr); 2358 ierr = MPI_Comm_rank(((PetscObject)A)->comm,&rank);CHKERRQ(ierr); 2359 2360 bs = A->rmap.bs; 2361 mbs = a->mbs; 2362 Mbs = a->Mbs; 2363 ba = b->a; 2364 bi = b->i; 2365 bj = b->j; 2366 2367 /* find ownerships */ 2368 rowners_bs = A->rmap.range; 2369 2370 /* each proc creates an array to be distributed */ 2371 ierr = PetscMalloc(bs*Mbs*sizeof(PetscReal),&work);CHKERRQ(ierr); 2372 ierr = PetscMemzero(work,bs*Mbs*sizeof(PetscReal));CHKERRQ(ierr); 2373 2374 /* row_max for B */ 2375 if (rank != size-1){ 2376 for (i=0; i<mbs; i++) { 2377 ncols = bi[1] - bi[0]; bi++; 2378 brow = bs*i; 2379 for (j=0; j<ncols; j++){ 2380 bcol = bs*(*bj); 2381 for (kcol=0; kcol<bs; kcol++){ 2382 col = bcol + kcol; /* local col index */ 2383 col += rowners_bs[rank+1]; /* global col index */ 2384 for (krow=0; krow<bs; krow++){ 2385 atmp = PetscAbsScalar(*ba); ba++; 2386 row = brow + krow; /* local row index */ 2387 if (PetscRealPart(va[row]) < atmp) va[row] = atmp; 2388 if (work[col] < atmp) work[col] = atmp; 2389 } 2390 } 2391 bj++; 2392 } 2393 } 2394 2395 /* send values to its owners */ 2396 for (dest=rank+1; dest<size; dest++){ 2397 svalues = work + rowners_bs[dest]; 2398 count = rowners_bs[dest+1]-rowners_bs[dest]; 2399 ierr = MPI_Send(svalues,count,MPIU_REAL,dest,rank,((PetscObject)A)->comm);CHKERRQ(ierr); 2400 } 2401 } 2402 2403 /* receive values */ 2404 if (rank){ 2405 rvalues = work; 2406 count = rowners_bs[rank+1]-rowners_bs[rank]; 2407 for (source=0; source<rank; source++){ 2408 ierr = MPI_Recv(rvalues,count,MPIU_REAL,MPI_ANY_SOURCE,MPI_ANY_TAG,((PetscObject)A)->comm,&stat);CHKERRQ(ierr); 2409 /* process values */ 2410 for (i=0; i<count; i++){ 2411 if (PetscRealPart(va[i]) < rvalues[i]) va[i] = rvalues[i]; 2412 } 2413 } 2414 } 2415 2416 ierr = VecRestoreArray(v,&va);CHKERRQ(ierr); 2417 ierr = PetscFree(work);CHKERRQ(ierr); 2418 PetscFunctionReturn(0); 2419 } 2420 2421 #undef __FUNCT__ 2422 #define __FUNCT__ "MatRelax_MPISBAIJ" 2423 PetscErrorCode MatRelax_MPISBAIJ(Mat matin,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx) 2424 { 2425 Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data; 2426 PetscErrorCode ierr; 2427 PetscInt mbs=mat->mbs,bs=matin->rmap.bs; 2428 PetscScalar *x,*b,*ptr,zero=0.0; 2429 Vec bb1; 2430 2431 PetscFunctionBegin; 2432 if (its <= 0 || lits <= 0) SETERRQ2(PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D and local its %D both positive",its,lits); 2433 if (bs > 1) 2434 SETERRQ(PETSC_ERR_SUP,"SSOR for block size > 1 is not yet implemented"); 2435 2436 if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP){ 2437 if ( flag & SOR_ZERO_INITIAL_GUESS ) { 2438 ierr = (*mat->A->ops->relax)(mat->A,bb,omega,flag,fshift,lits,lits,xx);CHKERRQ(ierr); 2439 its--; 2440 } 2441 2442 ierr = VecDuplicate(bb,&bb1);CHKERRQ(ierr); 2443 while (its--){ 2444 2445 /* lower triangular part: slvec0b = - B^T*xx */ 2446 ierr = (*mat->B->ops->multtranspose)(mat->B,xx,mat->slvec0b);CHKERRQ(ierr); 2447 2448 /* copy xx into slvec0a */ 2449 ierr = VecGetArray(mat->slvec0,&ptr);CHKERRQ(ierr); 2450 ierr = VecGetArray(xx,&x);CHKERRQ(ierr); 2451 ierr = PetscMemcpy(ptr,x,bs*mbs*sizeof(MatScalar));CHKERRQ(ierr); 2452 ierr = VecRestoreArray(mat->slvec0,&ptr);CHKERRQ(ierr); 2453 2454 ierr = VecScale(mat->slvec0,-1.0);CHKERRQ(ierr); 2455 2456 /* copy bb into slvec1a */ 2457 ierr = VecGetArray(mat->slvec1,&ptr);CHKERRQ(ierr); 2458 ierr = VecGetArray(bb,&b);CHKERRQ(ierr); 2459 ierr = PetscMemcpy(ptr,b,bs*mbs*sizeof(MatScalar));CHKERRQ(ierr); 2460 ierr = VecRestoreArray(mat->slvec1,&ptr);CHKERRQ(ierr); 2461 2462 /* set slvec1b = 0 */ 2463 ierr = VecSet(mat->slvec1b,zero);CHKERRQ(ierr); 2464 2465 ierr = VecScatterBegin(mat->sMvctx,mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2466 ierr = VecRestoreArray(xx,&x);CHKERRQ(ierr); 2467 ierr = VecRestoreArray(bb,&b);CHKERRQ(ierr); 2468 ierr = VecScatterEnd(mat->sMvctx,mat->slvec0,mat->slvec1,ADD_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2469 2470 /* upper triangular part: bb1 = bb1 - B*x */ 2471 ierr = (*mat->B->ops->multadd)(mat->B,mat->slvec1b,mat->slvec1a,bb1);CHKERRQ(ierr); 2472 2473 /* local diagonal sweep */ 2474 ierr = (*mat->A->ops->relax)(mat->A,bb1,omega,SOR_SYMMETRIC_SWEEP,fshift,lits,lits,xx);CHKERRQ(ierr); 2475 } 2476 ierr = VecDestroy(bb1);CHKERRQ(ierr); 2477 } else { 2478 SETERRQ(PETSC_ERR_SUP,"MatSORType is not supported for SBAIJ matrix format"); 2479 } 2480 PetscFunctionReturn(0); 2481 } 2482 2483 #undef __FUNCT__ 2484 #define __FUNCT__ "MatRelax_MPISBAIJ_2comm" 2485 PetscErrorCode MatRelax_MPISBAIJ_2comm(Mat matin,Vec bb,PetscReal omega,MatSORType flag,PetscReal fshift,PetscInt its,PetscInt lits,Vec xx) 2486 { 2487 Mat_MPISBAIJ *mat = (Mat_MPISBAIJ*)matin->data; 2488 PetscErrorCode ierr; 2489 Vec lvec1,bb1; 2490 2491 PetscFunctionBegin; 2492 if (its <= 0 || lits <= 0) SETERRQ2(PETSC_ERR_ARG_WRONG,"Relaxation requires global its %D and local its %D both positive",its,lits); 2493 if (matin->rmap.bs > 1) 2494 SETERRQ(PETSC_ERR_SUP,"SSOR for block size > 1 is not yet implemented"); 2495 2496 if ((flag & SOR_LOCAL_SYMMETRIC_SWEEP) == SOR_LOCAL_SYMMETRIC_SWEEP){ 2497 if ( flag & SOR_ZERO_INITIAL_GUESS ) { 2498 ierr = (*mat->A->ops->relax)(mat->A,bb,omega,flag,fshift,lits,lits,xx);CHKERRQ(ierr); 2499 its--; 2500 } 2501 2502 ierr = VecDuplicate(mat->lvec,&lvec1);CHKERRQ(ierr); 2503 ierr = VecDuplicate(bb,&bb1);CHKERRQ(ierr); 2504 while (its--){ 2505 ierr = VecScatterBegin(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2506 2507 /* lower diagonal part: bb1 = bb - B^T*xx */ 2508 ierr = (*mat->B->ops->multtranspose)(mat->B,xx,lvec1);CHKERRQ(ierr); 2509 ierr = VecScale(lvec1,-1.0);CHKERRQ(ierr); 2510 2511 ierr = VecScatterEnd(mat->Mvctx,xx,mat->lvec,INSERT_VALUES,SCATTER_FORWARD);CHKERRQ(ierr); 2512 ierr = VecCopy(bb,bb1);CHKERRQ(ierr); 2513 ierr = VecScatterBegin(mat->Mvctx,lvec1,bb1,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 2514 2515 /* upper diagonal part: bb1 = bb1 - B*x */ 2516 ierr = VecScale(mat->lvec,-1.0);CHKERRQ(ierr); 2517 ierr = (*mat->B->ops->multadd)(mat->B,mat->lvec,bb1,bb1);CHKERRQ(ierr); 2518 2519 ierr = VecScatterEnd(mat->Mvctx,lvec1,bb1,ADD_VALUES,SCATTER_REVERSE);CHKERRQ(ierr); 2520 2521 /* diagonal sweep */ 2522 ierr = (*mat->A->ops->relax)(mat->A,bb1,omega,SOR_SYMMETRIC_SWEEP,fshift,lits,lits,xx);CHKERRQ(ierr); 2523 } 2524 ierr = VecDestroy(lvec1);CHKERRQ(ierr); 2525 ierr = VecDestroy(bb1);CHKERRQ(ierr); 2526 } else { 2527 SETERRQ(PETSC_ERR_SUP,"MatSORType is not supported for SBAIJ matrix format"); 2528 } 2529 PetscFunctionReturn(0); 2530 } 2531 2532