1 /* Using Modified Sparse Row (MSR) storage. 2 See page 85, "Iterative Methods ..." by Saad. */ 3 4 /*$Id: sbaijfact.c,v 1.2 2000/06/23 22:00:05 buschelm Exp balay $*/ 5 /* 6 Factorization code for SBAIJ format. 7 */ 8 #include "sbaij.h" 9 #include "src/mat/impls/baij/seq/baij.h" 10 #include "src/vec/vecimpl.h" 11 #include "src/inline/ilu.h" 12 13 #undef __FUNC__ 14 #define __FUNC__ "MatLUFactorSymbolic_SeqSBAIJ" 15 int MatLUFactorSymbolic_SeqSBAIJ(Mat A,IS isrow,IS iscol,MatLUInfo *info,Mat *B) 16 { 17 Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data,*b; 18 IS isicol; 19 int *rip,*riip,ierr,i,mbs = a->mbs,*ai = a->i,*aj = a->j; 20 int *jutmp,bs = a->bs,bs2=a->bs2; 21 int m,nzi,realloc = 0; 22 int *jl,*q,jumin,jmin,jmax,juptr,nzk,qm,*iu,*ju,k,j,vj,umax,maxadd; 23 PetscReal f = 1.0; 24 25 PetscFunctionBegin; 26 PetscValidHeaderSpecific(isrow,IS_COOKIE); 27 PetscValidHeaderSpecific(iscol,IS_COOKIE); 28 /* if (A->M != A->N) SETERRQ(PETSC_ERR_ARG_WRONG,0,"matrix must be square");*/ 29 ierr = ISInvertPermutation(iscol,PETSC_DECIDE,&isicol);CHKERRQ(ierr); 30 ierr = ISGetIndices(isrow,&rip);CHKERRQ(ierr); 31 ierr = ISGetIndices(isicol,&riip);CHKERRQ(ierr); 32 for (k=0; k<mbs; k++) { 33 if ( rip[k] - riip[k] != 0 ) { 34 printf("Non-symm. permutation, use symm. permutation or general matrix format\n"); 35 break; 36 } 37 } 38 39 /* initialization */ 40 /* Don't know how many column pointers are needed so estimate. 41 Use Modified Sparse Row storage for u and ju, see Sasd pp.85 */ 42 if (info) f = info->fill; 43 umax = (int)(f*ai[mbs] + 1); umax += mbs + 1; 44 ju = iu = (int*)PetscMalloc(umax*sizeof(int));CHKPTRQ(ju); 45 iu[0] = mbs+1; 46 juptr = mbs; 47 jl = (int*)PetscMalloc(mbs*sizeof(int));CHKPTRQ(jl); 48 q = (int*)PetscMalloc(mbs*sizeof(int));CHKPTRQ(q); 49 for (i=0; i<mbs; i++){ 50 jl[i] = mbs; q[i] = 0; 51 } 52 53 /* for each row k */ 54 for (k=0; k<mbs; k++){ 55 nzk = 0; /* num. of nz blocks in k-th block row with diagonal block excluded */ 56 q[k] = mbs; 57 /* initialize nonzero structure of k-th row to row rip[k] of A */ 58 jmin = ai[rip[k]]; 59 jmax = ai[rip[k]+1]; 60 for (j=jmin; j<jmax; j++){ 61 vj = riip[aj[j]]; /* col. value */ 62 if(vj > k){ 63 qm = k; 64 do { 65 m = qm; qm = q[m]; 66 } while(qm < vj); 67 if (qm == vj) { 68 printf(" error: duplicate entry in A\n"); break; 69 } 70 nzk++; 71 q[m] = vj; 72 q[vj] = qm; 73 } /* if(vj > k) */ 74 } /* for (j=jmin; j<jmax; j++) */ 75 76 /* modify nonzero structure of k-th row by computing fill-in 77 for each row i to be merged in */ 78 i = k; 79 i = jl[i]; /* next pivot row (== mbs for symbolic factorization) */ 80 /* printf(" next pivot row i=%d\n",i); */ 81 while (i < mbs){ 82 /* merge row i into k-th row */ 83 nzi = iu[i+1] - (iu[i]+1); 84 jmin = iu[i] + 1; jmax = iu[i] + nzi; 85 qm = k; 86 for (j=jmin; j<jmax+1; j++){ 87 vj = ju[j]; 88 do { 89 m = qm; qm = q[m]; 90 } while (qm < vj); 91 if (qm != vj){ 92 nzk++; q[m] = vj; q[vj] = qm; qm = vj; 93 } 94 } 95 i = jl[i]; /* next pivot row */ 96 } 97 98 /* add k to row list for first nonzero element in k-th row */ 99 if (nzk > 0){ 100 i = q[k]; /* col value of first nonzero element in U(k, k+1:mbs-1) */ 101 jl[k] = jl[i]; jl[i] = k; 102 } 103 iu[k+1] = iu[k] + nzk; /* printf(" iu[%d]=%d, umax=%d\n", k+1, iu[k+1],umax);*/ 104 105 /* allocate more space to ju if needed */ 106 if (iu[k+1] > umax) { printf("allocate more space, iu[%d]=%d > umax=%d\n",k+1, iu[k+1],umax); 107 /* estimate how much additional space we will need */ 108 /* use the strategy suggested by David Hysom <hysom@perch-t.icase.edu> */ 109 /* just double the memory each time */ 110 maxadd = umax; 111 if (maxadd < nzk) maxadd = (mbs-k)*(nzk+1)/2; 112 umax += maxadd; 113 114 /* allocate a longer ju (NOTE: iu poits to the beginning of ju) */ 115 jutmp = (int*)PetscMalloc(umax*sizeof(int));CHKPTRQ(jutmp); 116 ierr = PetscMemcpy(jutmp,ju,iu[k]*sizeof(int));CHKERRQ(ierr); 117 ierr = PetscFree(ju);CHKERRQ(ierr); 118 ju = iu = jutmp; 119 realloc++; /* count how many times we realloc */ 120 } 121 122 /* save nonzero structure of k-th row in ju */ 123 i=k; 124 jumin = juptr + 1; juptr += nzk; 125 for (j=jumin; j<juptr+1; j++){ 126 i=q[i]; 127 ju[j]=i; 128 /* printf(" k=%d, ju[%d]=%d\n",k,j,ju[j]);*/ 129 } 130 /* printf("\n"); */ 131 } /* for (k=0; k<mbs; k++) */ 132 133 if (ai[mbs] != 0) { 134 PetscReal af = ((PetscReal)iu[mbs])/((PetscReal)ai[mbs]); 135 PLogInfo(A,"MatLUFactorSymbolic_SeqSBAIJ:Reallocs %d Fill ratio:given %g needed %g\n",realloc,f,af); 136 PLogInfo(A,"MatLUFactorSymbolic_SeqSBAIJ:Run with -pc_lu_fill %g or use \n",af); 137 PLogInfo(A,"MatLUFactorSymbolic_SeqSBAIJ:PCLUSetFill(pc,%g);\n",af); 138 PLogInfo(A,"MatLUFactorSymbolic_SeqSBAIJ:for best performance.\n"); 139 } else { 140 PLogInfo(A,"MatLUFactorSymbolic_SeqSBAIJ:Empty matrix.\n"); 141 } 142 143 ierr = ISRestoreIndices(isrow,&rip);CHKERRQ(ierr); 144 ierr = ISRestoreIndices(isicol,&riip);CHKERRQ(ierr); 145 146 ierr = PetscFree(q);CHKERRQ(ierr); 147 ierr = PetscFree(jl);CHKERRQ(ierr); 148 149 /* put together the new matrix */ 150 ierr = MatCreateSeqSBAIJ(A->comm,bs,bs*mbs,bs*mbs,0,PETSC_NULL,B);CHKERRQ(ierr); 151 PLogObjectParent(*B,isicol); 152 b = (Mat_SeqSBAIJ*)(*B)->data; 153 ierr = PetscFree(b->imax);CHKERRQ(ierr); 154 b->singlemalloc = PETSC_FALSE; 155 /* the next line frees the default space generated by the Create() */ 156 ierr = PetscFree(b->a);CHKERRQ(ierr); 157 ierr = PetscFree(b->ilen);CHKERRQ(ierr); 158 b->a = (MatScalar*)PetscMalloc((iu[mbs]+1)*sizeof(MatScalar)*bs2);CHKPTRQ(b->a); 159 b->j = ju; 160 b->i = iu; 161 b->diag = 0; 162 b->ilen = 0; 163 b->imax = 0; 164 b->row = isrow; 165 b->col = iscol; 166 ierr = PetscObjectReference((PetscObject)isrow);CHKERRQ(ierr); 167 ierr = PetscObjectReference((PetscObject)iscol);CHKERRQ(ierr); 168 b->icol = isicol; 169 b->solve_work = (Scalar*)PetscMalloc((bs*mbs+bs)*sizeof(Scalar));CHKPTRQ(b->solve_work); 170 /* In b structure: Free imax, ilen, old a, old j. 171 Allocate idnew, solve_work, new a, new j */ 172 PLogObjectMemory(*B,(iu[mbs]-mbs)*(sizeof(int)+sizeof(MatScalar))); 173 b->s_maxnz = b->s_nz = iu[mbs]; 174 175 (*B)->factor = FACTOR_LU; 176 (*B)->info.factor_mallocs = realloc; 177 (*B)->info.fill_ratio_given = f; 178 if (ai[mbs] != 0) { 179 (*B)->info.fill_ratio_needed = ((PetscReal)iu[mbs])/((PetscReal)ai[mbs]); 180 } else { 181 (*B)->info.fill_ratio_needed = 0.0; 182 } 183 184 185 PetscFunctionReturn(0); 186 } 187 188 /* ----------------------------------------------------------- */ 189 #undef __FUNC__ 190 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_N" 191 int MatLUFactorNumeric_SeqSBAIJ_N(Mat A,Mat *B) 192 { 193 Mat C = *B; 194 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 195 IS isrow = b->row,isicol = b->icol; 196 int *r,*ic,ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 197 int *ajtmpold,*ajtmp,nz,row,bslog,*ai=a->i,*aj=a->j,k,flg; 198 int *diag_offset=b->diag,diag,bs=a->bs,bs2 = a->bs2,*v_pivots,*pj; 199 MatScalar *ba = b->a,*aa = a->a,*pv,*v,*rtmp,*multiplier,*v_work,*pc,*w; 200 201 PetscFunctionBegin; 202 ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr); 203 ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr); 204 rtmp = (MatScalar*)PetscMalloc(bs2*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 205 ierr = PetscMemzero(rtmp,bs2*(n+1)*sizeof(MatScalar));CHKERRQ(ierr); 206 /* generate work space needed by dense LU factorization */ 207 v_work = (MatScalar*)PetscMalloc(bs*sizeof(int) + (bs+bs2)*sizeof(MatScalar));CHKPTRQ(v_work); 208 multiplier = v_work + bs; 209 v_pivots = (int*)(multiplier + bs2); 210 211 /* flops in while loop */ 212 bslog = 2*bs*bs2; 213 214 for (i=0; i<n; i++) { 215 nz = bi[i+1] - bi[i]; 216 ajtmp = bj + bi[i]; 217 for (j=0; j<nz; j++) { 218 ierr = PetscMemzero(rtmp+bs2*ajtmp[j],bs2*sizeof(MatScalar));CHKERRQ(ierr); 219 } 220 /* load in initial (unfactored row) */ 221 nz = ai[r[i]+1] - ai[r[i]]; 222 ajtmpold = aj + ai[r[i]]; 223 v = aa + bs2*ai[r[i]]; 224 for (j=0; j<nz; j++) { 225 ierr = PetscMemcpy(rtmp+bs2*ic[ajtmpold[j]],v+bs2*j,bs2*sizeof(MatScalar));CHKERRQ(ierr); 226 } 227 row = *ajtmp++; 228 while (row < i) { 229 pc = rtmp + bs2*row; 230 /* if (*pc) { */ 231 for (flg=0,k=0; k<bs2; k++) { if (pc[k]!=0.0) { flg =1; break; }} 232 if (flg) { 233 pv = ba + bs2*diag_offset[row]; 234 pj = bj + diag_offset[row] + 1; 235 Kernel_A_gets_A_times_B(bs,pc,pv,multiplier); 236 nz = bi[row+1] - diag_offset[row] - 1; 237 pv += bs2; 238 for (j=0; j<nz; j++) { 239 Kernel_A_gets_A_minus_B_times_C(bs,rtmp+bs2*pj[j],pc,pv+bs2*j); 240 } 241 PLogFlops(bslog*(nz+1)-bs); 242 } 243 row = *ajtmp++; 244 } 245 /* finished row so stick it into b->a */ 246 pv = ba + bs2*bi[i]; 247 pj = bj + bi[i]; 248 nz = bi[i+1] - bi[i]; 249 for (j=0; j<nz; j++) { 250 ierr = PetscMemcpy(pv+bs2*j,rtmp+bs2*pj[j],bs2*sizeof(MatScalar));CHKERRQ(ierr); 251 } 252 diag = diag_offset[i] - bi[i]; 253 /* invert diagonal block */ 254 w = pv + bs2*diag; 255 Kernel_A_gets_inverse_A(bs,w,v_pivots,v_work); 256 } 257 258 ierr = PetscFree(rtmp);CHKERRQ(ierr); 259 ierr = PetscFree(v_work);CHKERRQ(ierr); 260 ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr); 261 ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr); 262 C->factor = FACTOR_LU; 263 C->assembled = PETSC_TRUE; 264 PLogFlops(1.3333*bs*bs2*b->mbs); /* from inverting diagonal blocks */ 265 PetscFunctionReturn(0); 266 } 267 /* ------------------------------------------------------------*/ 268 /* 269 Version for when blocks are 7 by 7 270 */ 271 #undef __FUNC__ 272 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_7" 273 int MatLUFactorNumeric_SeqSBAIJ_7(Mat A,Mat *B) 274 { 275 Mat C = *B; 276 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 277 IS isrow = b->row,isicol = b->icol; 278 int *r,*ic,ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 279 int *ajtmpold,*ajtmp,nz,row; 280 int *diag_offset = b->diag,idx,*ai=a->i,*aj=a->j,*pj; 281 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 282 MatScalar p1,p2,p3,p4,m1,m2,m3,m4,m5,m6,m7,m8,m9,x1,x2,x3,x4; 283 MatScalar p5,p6,p7,p8,p9,x5,x6,x7,x8,x9,x10,x11,x12,x13,x14,x15,x16; 284 MatScalar x17,x18,x19,x20,x21,x22,x23,x24,x25,p10,p11,p12,p13,p14; 285 MatScalar p15,p16,p17,p18,p19,p20,p21,p22,p23,p24,p25,m10,m11,m12; 286 MatScalar m13,m14,m15,m16,m17,m18,m19,m20,m21,m22,m23,m24,m25; 287 MatScalar p26,p27,p28,p29,p30,p31,p32,p33,p34,p35,p36; 288 MatScalar p37,p38,p39,p40,p41,p42,p43,p44,p45,p46,p47,p48,p49; 289 MatScalar x26,x27,x28,x29,x30,x31,x32,x33,x34,x35,x36; 290 MatScalar x37,x38,x39,x40,x41,x42,x43,x44,x45,x46,x47,x48,x49; 291 MatScalar m26,m27,m28,m29,m30,m31,m32,m33,m34,m35,m36; 292 MatScalar m37,m38,m39,m40,m41,m42,m43,m44,m45,m46,m47,m48,m49; 293 MatScalar *ba = b->a,*aa = a->a; 294 295 PetscFunctionBegin; 296 ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr); 297 ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr); 298 rtmp = (MatScalar*)PetscMalloc(49*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 299 300 for (i=0; i<n; i++) { 301 nz = bi[i+1] - bi[i]; 302 ajtmp = bj + bi[i]; 303 for (j=0; j<nz; j++) { 304 x = rtmp+49*ajtmp[j]; 305 x[0] = x[1] = x[2] = x[3] = x[4] = x[5] = x[6] = x[7] = x[8] = x[9] = 0.0; 306 x[10] = x[11] = x[12] = x[13] = x[14] = x[15] = x[16] = x[17] = 0.0; 307 x[18] = x[19] = x[20] = x[21] = x[22] = x[23] = x[24] = x[25] = 0.0 ; 308 x[26] = x[27] = x[28] = x[29] = x[30] = x[31] = x[32] = x[33] = 0.0 ; 309 x[34] = x[35] = x[36] = x[37] = x[38] = x[39] = x[40] = x[41] = 0.0 ; 310 x[42] = x[43] = x[44] = x[45] = x[46] = x[47] = x[48] = 0.0 ; 311 } 312 /* load in initial (unfactored row) */ 313 idx = r[i]; 314 nz = ai[idx+1] - ai[idx]; 315 ajtmpold = aj + ai[idx]; 316 v = aa + 49*ai[idx]; 317 for (j=0; j<nz; j++) { 318 x = rtmp+49*ic[ajtmpold[j]]; 319 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 320 x[4] = v[4]; x[5] = v[5]; x[6] = v[6]; x[7] = v[7]; 321 x[8] = v[8]; x[9] = v[9]; x[10] = v[10]; x[11] = v[11]; 322 x[12] = v[12]; x[13] = v[13]; x[14] = v[14]; x[15] = v[15]; 323 x[16] = v[16]; x[17] = v[17]; x[18] = v[18]; x[19] = v[19]; 324 x[20] = v[20]; x[21] = v[21]; x[22] = v[22]; x[23] = v[23]; 325 x[24] = v[24]; x[25] = v[25]; x[26] = v[26]; x[27] = v[27]; 326 x[28] = v[28]; x[29] = v[29]; x[30] = v[30]; x[31] = v[31]; 327 x[32] = v[32]; x[33] = v[33]; x[34] = v[34]; x[35] = v[35]; 328 x[36] = v[36]; x[37] = v[37]; x[38] = v[38]; x[39] = v[39]; 329 x[40] = v[40]; x[41] = v[41]; x[42] = v[42]; x[43] = v[43]; 330 x[44] = v[44]; x[45] = v[45]; x[46] = v[46]; x[47] = v[47]; 331 x[48] = v[48]; 332 v += 49; 333 } 334 row = *ajtmp++; 335 while (row < i) { 336 pc = rtmp + 49*row; 337 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 338 p5 = pc[4]; p6 = pc[5]; p7 = pc[6]; p8 = pc[7]; 339 p9 = pc[8]; p10 = pc[9]; p11 = pc[10]; p12 = pc[11]; 340 p13 = pc[12]; p14 = pc[13]; p15 = pc[14]; p16 = pc[15]; 341 p17 = pc[16]; p18 = pc[17]; p19 = pc[18]; p20 = pc[19]; 342 p21 = pc[20]; p22 = pc[21]; p23 = pc[22]; p24 = pc[23]; 343 p25 = pc[24]; p26 = pc[25]; p27 = pc[26]; p28 = pc[27]; 344 p29 = pc[28]; p30 = pc[29]; p31 = pc[30]; p32 = pc[31]; 345 p33 = pc[32]; p34 = pc[33]; p35 = pc[34]; p36 = pc[35]; 346 p37 = pc[36]; p38 = pc[37]; p39 = pc[38]; p40 = pc[39]; 347 p41 = pc[40]; p42 = pc[41]; p43 = pc[42]; p44 = pc[43]; 348 p45 = pc[44]; p46 = pc[45]; p47 = pc[46]; p48 = pc[47]; 349 p49 = pc[48]; 350 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0 || 351 p5 != 0.0 || p6 != 0.0 || p7 != 0.0 || p8 != 0.0 || 352 p9 != 0.0 || p10 != 0.0 || p11 != 0.0 || p12 != 0.0 || 353 p13 != 0.0 || p14 != 0.0 || p15 != 0.0 || p16 != 0.0 || 354 p17 != 0.0 || p18 != 0.0 || p19 != 0.0 || p20 != 0.0 || 355 p21 != 0.0 || p22 != 0.0 || p23 != 0.0 || p24 != 0.0 || 356 p25 != 0.0 || p26 != 0.0 || p27 != 0.0 || p28 != 0.0 || 357 p29 != 0.0 || p30 != 0.0 || p31 != 0.0 || p32 != 0.0 || 358 p33 != 0.0 || p34 != 0.0 || p35 != 0.0 || p36 != 0.0 || 359 p37 != 0.0 || p38 != 0.0 || p39 != 0.0 || p40 != 0.0 || 360 p41 != 0.0 || p42 != 0.0 || p43 != 0.0 || p44 != 0.0 || 361 p45 != 0.0 || p46 != 0.0 || p47 != 0.0 || p48 != 0.0 || 362 p49 != 0.0) { 363 pv = ba + 49*diag_offset[row]; 364 pj = bj + diag_offset[row] + 1; 365 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 366 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; 367 x9 = pv[8]; x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; 368 x13 = pv[12]; x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; 369 x17 = pv[16]; x18 = pv[17]; x19 = pv[18]; x20 = pv[19]; 370 x21 = pv[20]; x22 = pv[21]; x23 = pv[22]; x24 = pv[23]; 371 x25 = pv[24]; x26 = pv[25]; x27 = pv[26]; x28 = pv[27]; 372 x29 = pv[28]; x30 = pv[29]; x31 = pv[30]; x32 = pv[31]; 373 x33 = pv[32]; x34 = pv[33]; x35 = pv[34]; x36 = pv[35]; 374 x37 = pv[36]; x38 = pv[37]; x39 = pv[38]; x40 = pv[39]; 375 x41 = pv[40]; x42 = pv[41]; x43 = pv[42]; x44 = pv[43]; 376 x45 = pv[44]; x46 = pv[45]; x47 = pv[46]; x48 = pv[47]; 377 x49 = pv[48]; 378 pc[0] = m1 = p1*x1 + p8*x2 + p15*x3 + p22*x4 + p29*x5 + p36*x6 + p43*x7; 379 pc[1] = m2 = p2*x1 + p9*x2 + p16*x3 + p23*x4 + p30*x5 + p37*x6 + p44*x7; 380 pc[2] = m3 = p3*x1 + p10*x2 + p17*x3 + p24*x4 + p31*x5 + p38*x6 + p45*x7; 381 pc[3] = m4 = p4*x1 + p11*x2 + p18*x3 + p25*x4 + p32*x5 + p39*x6 + p46*x7; 382 pc[4] = m5 = p5*x1 + p12*x2 + p19*x3 + p26*x4 + p33*x5 + p40*x6 + p47*x7; 383 pc[5] = m6 = p6*x1 + p13*x2 + p20*x3 + p27*x4 + p34*x5 + p41*x6 + p48*x7; 384 pc[6] = m7 = p7*x1 + p14*x2 + p21*x3 + p28*x4 + p35*x5 + p42*x6 + p49*x7; 385 386 pc[7] = m8 = p1*x8 + p8*x9 + p15*x10 + p22*x11 + p29*x12 + p36*x13 + p43*x14; 387 pc[8] = m9 = p2*x8 + p9*x9 + p16*x10 + p23*x11 + p30*x12 + p37*x13 + p44*x14; 388 pc[9] = m10 = p3*x8 + p10*x9 + p17*x10 + p24*x11 + p31*x12 + p38*x13 + p45*x14; 389 pc[10] = m11 = p4*x8 + p11*x9 + p18*x10 + p25*x11 + p32*x12 + p39*x13 + p46*x14; 390 pc[11] = m12 = p5*x8 + p12*x9 + p19*x10 + p26*x11 + p33*x12 + p40*x13 + p47*x14; 391 pc[12] = m13 = p6*x8 + p13*x9 + p20*x10 + p27*x11 + p34*x12 + p41*x13 + p48*x14; 392 pc[13] = m14 = p7*x8 + p14*x9 + p21*x10 + p28*x11 + p35*x12 + p42*x13 + p49*x14; 393 394 pc[14] = m15 = p1*x15 + p8*x16 + p15*x17 + p22*x18 + p29*x19 + p36*x20 + p43*x21; 395 pc[15] = m16 = p2*x15 + p9*x16 + p16*x17 + p23*x18 + p30*x19 + p37*x20 + p44*x21; 396 pc[16] = m17 = p3*x15 + p10*x16 + p17*x17 + p24*x18 + p31*x19 + p38*x20 + p45*x21; 397 pc[17] = m18 = p4*x15 + p11*x16 + p18*x17 + p25*x18 + p32*x19 + p39*x20 + p46*x21; 398 pc[18] = m19 = p5*x15 + p12*x16 + p19*x17 + p26*x18 + p33*x19 + p40*x20 + p47*x21; 399 pc[19] = m20 = p6*x15 + p13*x16 + p20*x17 + p27*x18 + p34*x19 + p41*x20 + p48*x21; 400 pc[20] = m21 = p7*x15 + p14*x16 + p21*x17 + p28*x18 + p35*x19 + p42*x20 + p49*x21; 401 402 pc[21] = m22 = p1*x22 + p8*x23 + p15*x24 + p22*x25 + p29*x26 + p36*x27 + p43*x28; 403 pc[22] = m23 = p2*x22 + p9*x23 + p16*x24 + p23*x25 + p30*x26 + p37*x27 + p44*x28; 404 pc[23] = m24 = p3*x22 + p10*x23 + p17*x24 + p24*x25 + p31*x26 + p38*x27 + p45*x28; 405 pc[24] = m25 = p4*x22 + p11*x23 + p18*x24 + p25*x25 + p32*x26 + p39*x27 + p46*x28; 406 pc[25] = m26 = p5*x22 + p12*x23 + p19*x24 + p26*x25 + p33*x26 + p40*x27 + p47*x28; 407 pc[26] = m27 = p6*x22 + p13*x23 + p20*x24 + p27*x25 + p34*x26 + p41*x27 + p48*x28; 408 pc[27] = m28 = p7*x22 + p14*x23 + p21*x24 + p28*x25 + p35*x26 + p42*x27 + p49*x28; 409 410 pc[28] = m29 = p1*x29 + p8*x30 + p15*x31 + p22*x32 + p29*x33 + p36*x34 + p43*x35; 411 pc[29] = m30 = p2*x29 + p9*x30 + p16*x31 + p23*x32 + p30*x33 + p37*x34 + p44*x35; 412 pc[30] = m31 = p3*x29 + p10*x30 + p17*x31 + p24*x32 + p31*x33 + p38*x34 + p45*x35; 413 pc[31] = m32 = p4*x29 + p11*x30 + p18*x31 + p25*x32 + p32*x33 + p39*x34 + p46*x35; 414 pc[32] = m33 = p5*x29 + p12*x30 + p19*x31 + p26*x32 + p33*x33 + p40*x34 + p47*x35; 415 pc[33] = m34 = p6*x29 + p13*x30 + p20*x31 + p27*x32 + p34*x33 + p41*x34 + p48*x35; 416 pc[34] = m35 = p7*x29 + p14*x30 + p21*x31 + p28*x32 + p35*x33 + p42*x34 + p49*x35; 417 418 pc[35] = m36 = p1*x36 + p8*x37 + p15*x38 + p22*x39 + p29*x40 + p36*x41 + p43*x42; 419 pc[36] = m37 = p2*x36 + p9*x37 + p16*x38 + p23*x39 + p30*x40 + p37*x41 + p44*x42; 420 pc[37] = m38 = p3*x36 + p10*x37 + p17*x38 + p24*x39 + p31*x40 + p38*x41 + p45*x42; 421 pc[38] = m39 = p4*x36 + p11*x37 + p18*x38 + p25*x39 + p32*x40 + p39*x41 + p46*x42; 422 pc[39] = m40 = p5*x36 + p12*x37 + p19*x38 + p26*x39 + p33*x40 + p40*x41 + p47*x42; 423 pc[40] = m41 = p6*x36 + p13*x37 + p20*x38 + p27*x39 + p34*x40 + p41*x41 + p48*x42; 424 pc[41] = m42 = p7*x36 + p14*x37 + p21*x38 + p28*x39 + p35*x40 + p42*x41 + p49*x42; 425 426 pc[42] = m43 = p1*x43 + p8*x44 + p15*x45 + p22*x46 + p29*x47 + p36*x48 + p43*x49; 427 pc[43] = m44 = p2*x43 + p9*x44 + p16*x45 + p23*x46 + p30*x47 + p37*x48 + p44*x49; 428 pc[44] = m45 = p3*x43 + p10*x44 + p17*x45 + p24*x46 + p31*x47 + p38*x48 + p45*x49; 429 pc[45] = m46 = p4*x43 + p11*x44 + p18*x45 + p25*x46 + p32*x47 + p39*x48 + p46*x49; 430 pc[46] = m47 = p5*x43 + p12*x44 + p19*x45 + p26*x46 + p33*x47 + p40*x48 + p47*x49; 431 pc[47] = m48 = p6*x43 + p13*x44 + p20*x45 + p27*x46 + p34*x47 + p41*x48 + p48*x49; 432 pc[48] = m49 = p7*x43 + p14*x44 + p21*x45 + p28*x46 + p35*x47 + p42*x48 + p49*x49; 433 434 nz = bi[row+1] - diag_offset[row] - 1; 435 pv += 49; 436 for (j=0; j<nz; j++) { 437 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 438 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; 439 x9 = pv[8]; x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; 440 x13 = pv[12]; x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; 441 x17 = pv[16]; x18 = pv[17]; x19 = pv[18]; x20 = pv[19]; 442 x21 = pv[20]; x22 = pv[21]; x23 = pv[22]; x24 = pv[23]; 443 x25 = pv[24]; x26 = pv[25]; x27 = pv[26]; x28 = pv[27]; 444 x29 = pv[28]; x30 = pv[29]; x31 = pv[30]; x32 = pv[31]; 445 x33 = pv[32]; x34 = pv[33]; x35 = pv[34]; x36 = pv[35]; 446 x37 = pv[36]; x38 = pv[37]; x39 = pv[38]; x40 = pv[39]; 447 x41 = pv[40]; x42 = pv[41]; x43 = pv[42]; x44 = pv[43]; 448 x45 = pv[44]; x46 = pv[45]; x47 = pv[46]; x48 = pv[47]; 449 x49 = pv[48]; 450 x = rtmp + 49*pj[j]; 451 x[0] -= m1*x1 + m8*x2 + m15*x3 + m22*x4 + m29*x5 + m36*x6 + m43*x7; 452 x[1] -= m2*x1 + m9*x2 + m16*x3 + m23*x4 + m30*x5 + m37*x6 + m44*x7; 453 x[2] -= m3*x1 + m10*x2 + m17*x3 + m24*x4 + m31*x5 + m38*x6 + m45*x7; 454 x[3] -= m4*x1 + m11*x2 + m18*x3 + m25*x4 + m32*x5 + m39*x6 + m46*x7; 455 x[4] -= m5*x1 + m12*x2 + m19*x3 + m26*x4 + m33*x5 + m40*x6 + m47*x7; 456 x[5] -= m6*x1 + m13*x2 + m20*x3 + m27*x4 + m34*x5 + m41*x6 + m48*x7; 457 x[6] -= m7*x1 + m14*x2 + m21*x3 + m28*x4 + m35*x5 + m42*x6 + m49*x7; 458 459 x[7] -= m1*x8 + m8*x9 + m15*x10 + m22*x11 + m29*x12 + m36*x13 + m43*x14; 460 x[8] -= m2*x8 + m9*x9 + m16*x10 + m23*x11 + m30*x12 + m37*x13 + m44*x14; 461 x[9] -= m3*x8 + m10*x9 + m17*x10 + m24*x11 + m31*x12 + m38*x13 + m45*x14; 462 x[10] -= m4*x8 + m11*x9 + m18*x10 + m25*x11 + m32*x12 + m39*x13 + m46*x14; 463 x[11] -= m5*x8 + m12*x9 + m19*x10 + m26*x11 + m33*x12 + m40*x13 + m47*x14; 464 x[12] -= m6*x8 + m13*x9 + m20*x10 + m27*x11 + m34*x12 + m41*x13 + m48*x14; 465 x[13] -= m7*x8 + m14*x9 + m21*x10 + m28*x11 + m35*x12 + m42*x13 + m49*x14; 466 467 x[14] -= m1*x15 + m8*x16 + m15*x17 + m22*x18 + m29*x19 + m36*x20 + m43*x21; 468 x[15] -= m2*x15 + m9*x16 + m16*x17 + m23*x18 + m30*x19 + m37*x20 + m44*x21; 469 x[16] -= m3*x15 + m10*x16 + m17*x17 + m24*x18 + m31*x19 + m38*x20 + m45*x21; 470 x[17] -= m4*x15 + m11*x16 + m18*x17 + m25*x18 + m32*x19 + m39*x20 + m46*x21; 471 x[18] -= m5*x15 + m12*x16 + m19*x17 + m26*x18 + m33*x19 + m40*x20 + m47*x21; 472 x[19] -= m6*x15 + m13*x16 + m20*x17 + m27*x18 + m34*x19 + m41*x20 + m48*x21; 473 x[20] -= m7*x15 + m14*x16 + m21*x17 + m28*x18 + m35*x19 + m42*x20 + m49*x21; 474 475 x[21] -= m1*x22 + m8*x23 + m15*x24 + m22*x25 + m29*x26 + m36*x27 + m43*x28; 476 x[22] -= m2*x22 + m9*x23 + m16*x24 + m23*x25 + m30*x26 + m37*x27 + m44*x28; 477 x[23] -= m3*x22 + m10*x23 + m17*x24 + m24*x25 + m31*x26 + m38*x27 + m45*x28; 478 x[24] -= m4*x22 + m11*x23 + m18*x24 + m25*x25 + m32*x26 + m39*x27 + m46*x28; 479 x[25] -= m5*x22 + m12*x23 + m19*x24 + m26*x25 + m33*x26 + m40*x27 + m47*x28; 480 x[26] -= m6*x22 + m13*x23 + m20*x24 + m27*x25 + m34*x26 + m41*x27 + m48*x28; 481 x[27] -= m7*x22 + m14*x23 + m21*x24 + m28*x25 + m35*x26 + m42*x27 + m49*x28; 482 483 x[28] -= m1*x29 + m8*x30 + m15*x31 + m22*x32 + m29*x33 + m36*x34 + m43*x35; 484 x[29] -= m2*x29 + m9*x30 + m16*x31 + m23*x32 + m30*x33 + m37*x34 + m44*x35; 485 x[30] -= m3*x29 + m10*x30 + m17*x31 + m24*x32 + m31*x33 + m38*x34 + m45*x35; 486 x[31] -= m4*x29 + m11*x30 + m18*x31 + m25*x32 + m32*x33 + m39*x34 + m46*x35; 487 x[32] -= m5*x29 + m12*x30 + m19*x31 + m26*x32 + m33*x33 + m40*x34 + m47*x35; 488 x[33] -= m6*x29 + m13*x30 + m20*x31 + m27*x32 + m34*x33 + m41*x34 + m48*x35; 489 x[34] -= m7*x29 + m14*x30 + m21*x31 + m28*x32 + m35*x33 + m42*x34 + m49*x35; 490 491 x[35] -= m1*x36 + m8*x37 + m15*x38 + m22*x39 + m29*x40 + m36*x41 + m43*x42; 492 x[36] -= m2*x36 + m9*x37 + m16*x38 + m23*x39 + m30*x40 + m37*x41 + m44*x42; 493 x[37] -= m3*x36 + m10*x37 + m17*x38 + m24*x39 + m31*x40 + m38*x41 + m45*x42; 494 x[38] -= m4*x36 + m11*x37 + m18*x38 + m25*x39 + m32*x40 + m39*x41 + m46*x42; 495 x[39] -= m5*x36 + m12*x37 + m19*x38 + m26*x39 + m33*x40 + m40*x41 + m47*x42; 496 x[40] -= m6*x36 + m13*x37 + m20*x38 + m27*x39 + m34*x40 + m41*x41 + m48*x42; 497 x[41] -= m7*x36 + m14*x37 + m21*x38 + m28*x39 + m35*x40 + m42*x41 + m49*x42; 498 499 x[42] -= m1*x43 + m8*x44 + m15*x45 + m22*x46 + m29*x47 + m36*x48 + m43*x49; 500 x[43] -= m2*x43 + m9*x44 + m16*x45 + m23*x46 + m30*x47 + m37*x48 + m44*x49; 501 x[44] -= m3*x43 + m10*x44 + m17*x45 + m24*x46 + m31*x47 + m38*x48 + m45*x49; 502 x[45] -= m4*x43 + m11*x44 + m18*x45 + m25*x46 + m32*x47 + m39*x48 + m46*x49; 503 x[46] -= m5*x43 + m12*x44 + m19*x45 + m26*x46 + m33*x47 + m40*x48 + m47*x49; 504 x[47] -= m6*x43 + m13*x44 + m20*x45 + m27*x46 + m34*x47 + m41*x48 + m48*x49; 505 x[48] -= m7*x43 + m14*x44 + m21*x45 + m28*x46 + m35*x47 + m42*x48 + m49*x49; 506 pv += 49; 507 } 508 PLogFlops(686*nz+637); 509 } 510 row = *ajtmp++; 511 } 512 /* finished row so stick it into b->a */ 513 pv = ba + 49*bi[i]; 514 pj = bj + bi[i]; 515 nz = bi[i+1] - bi[i]; 516 for (j=0; j<nz; j++) { 517 x = rtmp+49*pj[j]; 518 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 519 pv[4] = x[4]; pv[5] = x[5]; pv[6] = x[6]; pv[7] = x[7]; 520 pv[8] = x[8]; pv[9] = x[9]; pv[10] = x[10]; pv[11] = x[11]; 521 pv[12] = x[12]; pv[13] = x[13]; pv[14] = x[14]; pv[15] = x[15]; 522 pv[16] = x[16]; pv[17] = x[17]; pv[18] = x[18]; pv[19] = x[19]; 523 pv[20] = x[20]; pv[21] = x[21]; pv[22] = x[22]; pv[23] = x[23]; 524 pv[24] = x[24]; pv[25] = x[25]; pv[26] = x[26]; pv[27] = x[27]; 525 pv[28] = x[28]; pv[29] = x[29]; pv[30] = x[30]; pv[31] = x[31]; 526 pv[32] = x[32]; pv[33] = x[33]; pv[34] = x[34]; pv[35] = x[35]; 527 pv[36] = x[36]; pv[37] = x[37]; pv[38] = x[38]; pv[39] = x[39]; 528 pv[40] = x[40]; pv[41] = x[41]; pv[42] = x[42]; pv[43] = x[43]; 529 pv[44] = x[44]; pv[45] = x[45]; pv[46] = x[46]; pv[47] = x[47]; 530 pv[48] = x[48]; 531 pv += 49; 532 } 533 /* invert diagonal block */ 534 w = ba + 49*diag_offset[i]; 535 ierr = Kernel_A_gets_inverse_A_7(w);CHKERRQ(ierr); 536 } 537 538 ierr = PetscFree(rtmp);CHKERRQ(ierr); 539 ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr); 540 ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr); 541 C->factor = FACTOR_LU; 542 C->assembled = PETSC_TRUE; 543 PLogFlops(1.3333*343*b->mbs); /* from inverting diagonal blocks */ 544 PetscFunctionReturn(0); 545 } 546 547 /* 548 Version for when blocks are 7 by 7 Using natural ordering 549 */ 550 #undef __FUNC__ 551 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_7_NaturalOrdering" 552 int MatLUFactorNumeric_SeqSBAIJ_7_NaturalOrdering(Mat A,Mat *B) 553 { 554 Mat C = *B; 555 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 556 int ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 557 int *ajtmpold,*ajtmp,nz,row; 558 int *diag_offset = b->diag,*ai=a->i,*aj=a->j,*pj; 559 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 560 MatScalar x1,x2,x3,x4,x5,x6,x7,x8,x9,x10,x11,x12,x13,x14,x15; 561 MatScalar x16,x17,x18,x19,x20,x21,x22,x23,x24,x25; 562 MatScalar p1,p2,p3,p4,p5,p6,p7,p8,p9,p10,p11,p12,p13,p14,p15; 563 MatScalar p16,p17,p18,p19,p20,p21,p22,p23,p24,p25; 564 MatScalar m1,m2,m3,m4,m5,m6,m7,m8,m9,m10,m11,m12,m13,m14,m15; 565 MatScalar m16,m17,m18,m19,m20,m21,m22,m23,m24,m25; 566 MatScalar p26,p27,p28,p29,p30,p31,p32,p33,p34,p35,p36; 567 MatScalar p37,p38,p39,p40,p41,p42,p43,p44,p45,p46,p47,p48,p49; 568 MatScalar x26,x27,x28,x29,x30,x31,x32,x33,x34,x35,x36; 569 MatScalar x37,x38,x39,x40,x41,x42,x43,x44,x45,x46,x47,x48,x49; 570 MatScalar m26,m27,m28,m29,m30,m31,m32,m33,m34,m35,m36; 571 MatScalar m37,m38,m39,m40,m41,m42,m43,m44,m45,m46,m47,m48,m49; 572 MatScalar *ba = b->a,*aa = a->a; 573 574 PetscFunctionBegin; 575 rtmp = (MatScalar*)PetscMalloc(49*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 576 for (i=0; i<n; i++) { 577 nz = bi[i+1] - bi[i]; 578 ajtmp = bj + bi[i]; 579 for (j=0; j<nz; j++) { 580 x = rtmp+49*ajtmp[j]; 581 x[0] = x[1] = x[2] = x[3] = x[4] = x[5] = x[6] = x[7] = x[8] = x[9] = 0.0; 582 x[10] = x[11] = x[12] = x[13] = x[14] = x[15] = x[16] = x[17] = 0.0; 583 x[18] = x[19] = x[20] = x[21] = x[22] = x[23] = x[24] = x[25] = 0.0 ; 584 x[26] = x[27] = x[28] = x[29] = x[30] = x[31] = x[32] = x[33] = 0.0 ; 585 x[34] = x[35] = x[36] = x[37] = x[38] = x[39] = x[40] = x[41] = 0.0 ; 586 x[42] = x[43] = x[44] = x[45] = x[46] = x[47] = x[48] = 0.0 ; 587 } 588 /* load in initial (unfactored row) */ 589 nz = ai[i+1] - ai[i]; 590 ajtmpold = aj + ai[i]; 591 v = aa + 49*ai[i]; 592 for (j=0; j<nz; j++) { 593 x = rtmp+49*ajtmpold[j]; 594 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 595 x[4] = v[4]; x[5] = v[5]; x[6] = v[6]; x[7] = v[7]; 596 x[8] = v[8]; x[9] = v[9]; x[10] = v[10]; x[11] = v[11]; 597 x[12] = v[12]; x[13] = v[13]; x[14] = v[14]; x[15] = v[15]; 598 x[16] = v[16]; x[17] = v[17]; x[18] = v[18]; x[19] = v[19]; 599 x[20] = v[20]; x[21] = v[21]; x[22] = v[22]; x[23] = v[23]; 600 x[24] = v[24]; x[25] = v[25]; x[26] = v[26]; x[27] = v[27]; 601 x[28] = v[28]; x[29] = v[29]; x[30] = v[30]; x[31] = v[31]; 602 x[32] = v[32]; x[33] = v[33]; x[34] = v[34]; x[35] = v[35]; 603 x[36] = v[36]; x[37] = v[37]; x[38] = v[38]; x[39] = v[39]; 604 x[40] = v[40]; x[41] = v[41]; x[42] = v[42]; x[43] = v[43]; 605 x[44] = v[44]; x[45] = v[45]; x[46] = v[46]; x[47] = v[47]; 606 x[48] = v[48]; 607 v += 49; 608 } 609 row = *ajtmp++; 610 while (row < i) { 611 pc = rtmp + 49*row; 612 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 613 p5 = pc[4]; p6 = pc[5]; p7 = pc[6]; p8 = pc[7]; 614 p9 = pc[8]; p10 = pc[9]; p11 = pc[10]; p12 = pc[11]; 615 p13 = pc[12]; p14 = pc[13]; p15 = pc[14]; p16 = pc[15]; 616 p17 = pc[16]; p18 = pc[17]; p19 = pc[18]; p20 = pc[19]; 617 p21 = pc[20]; p22 = pc[21]; p23 = pc[22]; p24 = pc[23]; 618 p25 = pc[24]; p26 = pc[25]; p27 = pc[26]; p28 = pc[27]; 619 p29 = pc[28]; p30 = pc[29]; p31 = pc[30]; p32 = pc[31]; 620 p33 = pc[32]; p34 = pc[33]; p35 = pc[34]; p36 = pc[35]; 621 p37 = pc[36]; p38 = pc[37]; p39 = pc[38]; p40 = pc[39]; 622 p41 = pc[40]; p42 = pc[41]; p43 = pc[42]; p44 = pc[43]; 623 p45 = pc[44]; p46 = pc[45]; p47 = pc[46]; p48 = pc[47]; 624 p49 = pc[48]; 625 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0 || 626 p5 != 0.0 || p6 != 0.0 || p7 != 0.0 || p8 != 0.0 || 627 p9 != 0.0 || p10 != 0.0 || p11 != 0.0 || p12 != 0.0 || 628 p13 != 0.0 || p14 != 0.0 || p15 != 0.0 || p16 != 0.0 || 629 p17 != 0.0 || p18 != 0.0 || p19 != 0.0 || p20 != 0.0 || 630 p21 != 0.0 || p22 != 0.0 || p23 != 0.0 || p24 != 0.0 || 631 p25 != 0.0 || p26 != 0.0 || p27 != 0.0 || p28 != 0.0 || 632 p29 != 0.0 || p30 != 0.0 || p31 != 0.0 || p32 != 0.0 || 633 p33 != 0.0 || p34 != 0.0 || p35 != 0.0 || p36 != 0.0 || 634 p37 != 0.0 || p38 != 0.0 || p39 != 0.0 || p40 != 0.0 || 635 p41 != 0.0 || p42 != 0.0 || p43 != 0.0 || p44 != 0.0 || 636 p45 != 0.0 || p46 != 0.0 || p47 != 0.0 || p48 != 0.0 || 637 p49 != 0.0) { 638 pv = ba + 49*diag_offset[row]; 639 pj = bj + diag_offset[row] + 1; 640 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 641 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; 642 x9 = pv[8]; x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; 643 x13 = pv[12]; x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; 644 x17 = pv[16]; x18 = pv[17]; x19 = pv[18]; x20 = pv[19]; 645 x21 = pv[20]; x22 = pv[21]; x23 = pv[22]; x24 = pv[23]; 646 x25 = pv[24]; x26 = pv[25]; x27 = pv[26]; x28 = pv[27]; 647 x29 = pv[28]; x30 = pv[29]; x31 = pv[30]; x32 = pv[31]; 648 x33 = pv[32]; x34 = pv[33]; x35 = pv[34]; x36 = pv[35]; 649 x37 = pv[36]; x38 = pv[37]; x39 = pv[38]; x40 = pv[39]; 650 x41 = pv[40]; x42 = pv[41]; x43 = pv[42]; x44 = pv[43]; 651 x45 = pv[44]; x46 = pv[45]; x47 = pv[46]; x48 = pv[47]; 652 x49 = pv[48]; 653 pc[0] = m1 = p1*x1 + p8*x2 + p15*x3 + p22*x4 + p29*x5 + p36*x6 + p43*x7; 654 pc[1] = m2 = p2*x1 + p9*x2 + p16*x3 + p23*x4 + p30*x5 + p37*x6 + p44*x7; 655 pc[2] = m3 = p3*x1 + p10*x2 + p17*x3 + p24*x4 + p31*x5 + p38*x6 + p45*x7; 656 pc[3] = m4 = p4*x1 + p11*x2 + p18*x3 + p25*x4 + p32*x5 + p39*x6 + p46*x7; 657 pc[4] = m5 = p5*x1 + p12*x2 + p19*x3 + p26*x4 + p33*x5 + p40*x6 + p47*x7; 658 pc[5] = m6 = p6*x1 + p13*x2 + p20*x3 + p27*x4 + p34*x5 + p41*x6 + p48*x7; 659 pc[6] = m7 = p7*x1 + p14*x2 + p21*x3 + p28*x4 + p35*x5 + p42*x6 + p49*x7; 660 661 pc[7] = m8 = p1*x8 + p8*x9 + p15*x10 + p22*x11 + p29*x12 + p36*x13 + p43*x14; 662 pc[8] = m9 = p2*x8 + p9*x9 + p16*x10 + p23*x11 + p30*x12 + p37*x13 + p44*x14; 663 pc[9] = m10 = p3*x8 + p10*x9 + p17*x10 + p24*x11 + p31*x12 + p38*x13 + p45*x14; 664 pc[10] = m11 = p4*x8 + p11*x9 + p18*x10 + p25*x11 + p32*x12 + p39*x13 + p46*x14; 665 pc[11] = m12 = p5*x8 + p12*x9 + p19*x10 + p26*x11 + p33*x12 + p40*x13 + p47*x14; 666 pc[12] = m13 = p6*x8 + p13*x9 + p20*x10 + p27*x11 + p34*x12 + p41*x13 + p48*x14; 667 pc[13] = m14 = p7*x8 + p14*x9 + p21*x10 + p28*x11 + p35*x12 + p42*x13 + p49*x14; 668 669 pc[14] = m15 = p1*x15 + p8*x16 + p15*x17 + p22*x18 + p29*x19 + p36*x20 + p43*x21; 670 pc[15] = m16 = p2*x15 + p9*x16 + p16*x17 + p23*x18 + p30*x19 + p37*x20 + p44*x21; 671 pc[16] = m17 = p3*x15 + p10*x16 + p17*x17 + p24*x18 + p31*x19 + p38*x20 + p45*x21; 672 pc[17] = m18 = p4*x15 + p11*x16 + p18*x17 + p25*x18 + p32*x19 + p39*x20 + p46*x21; 673 pc[18] = m19 = p5*x15 + p12*x16 + p19*x17 + p26*x18 + p33*x19 + p40*x20 + p47*x21; 674 pc[19] = m20 = p6*x15 + p13*x16 + p20*x17 + p27*x18 + p34*x19 + p41*x20 + p48*x21; 675 pc[20] = m21 = p7*x15 + p14*x16 + p21*x17 + p28*x18 + p35*x19 + p42*x20 + p49*x21; 676 677 pc[21] = m22 = p1*x22 + p8*x23 + p15*x24 + p22*x25 + p29*x26 + p36*x27 + p43*x28; 678 pc[22] = m23 = p2*x22 + p9*x23 + p16*x24 + p23*x25 + p30*x26 + p37*x27 + p44*x28; 679 pc[23] = m24 = p3*x22 + p10*x23 + p17*x24 + p24*x25 + p31*x26 + p38*x27 + p45*x28; 680 pc[24] = m25 = p4*x22 + p11*x23 + p18*x24 + p25*x25 + p32*x26 + p39*x27 + p46*x28; 681 pc[25] = m26 = p5*x22 + p12*x23 + p19*x24 + p26*x25 + p33*x26 + p40*x27 + p47*x28; 682 pc[26] = m27 = p6*x22 + p13*x23 + p20*x24 + p27*x25 + p34*x26 + p41*x27 + p48*x28; 683 pc[27] = m28 = p7*x22 + p14*x23 + p21*x24 + p28*x25 + p35*x26 + p42*x27 + p49*x28; 684 685 pc[28] = m29 = p1*x29 + p8*x30 + p15*x31 + p22*x32 + p29*x33 + p36*x34 + p43*x35; 686 pc[29] = m30 = p2*x29 + p9*x30 + p16*x31 + p23*x32 + p30*x33 + p37*x34 + p44*x35; 687 pc[30] = m31 = p3*x29 + p10*x30 + p17*x31 + p24*x32 + p31*x33 + p38*x34 + p45*x35; 688 pc[31] = m32 = p4*x29 + p11*x30 + p18*x31 + p25*x32 + p32*x33 + p39*x34 + p46*x35; 689 pc[32] = m33 = p5*x29 + p12*x30 + p19*x31 + p26*x32 + p33*x33 + p40*x34 + p47*x35; 690 pc[33] = m34 = p6*x29 + p13*x30 + p20*x31 + p27*x32 + p34*x33 + p41*x34 + p48*x35; 691 pc[34] = m35 = p7*x29 + p14*x30 + p21*x31 + p28*x32 + p35*x33 + p42*x34 + p49*x35; 692 693 pc[35] = m36 = p1*x36 + p8*x37 + p15*x38 + p22*x39 + p29*x40 + p36*x41 + p43*x42; 694 pc[36] = m37 = p2*x36 + p9*x37 + p16*x38 + p23*x39 + p30*x40 + p37*x41 + p44*x42; 695 pc[37] = m38 = p3*x36 + p10*x37 + p17*x38 + p24*x39 + p31*x40 + p38*x41 + p45*x42; 696 pc[38] = m39 = p4*x36 + p11*x37 + p18*x38 + p25*x39 + p32*x40 + p39*x41 + p46*x42; 697 pc[39] = m40 = p5*x36 + p12*x37 + p19*x38 + p26*x39 + p33*x40 + p40*x41 + p47*x42; 698 pc[40] = m41 = p6*x36 + p13*x37 + p20*x38 + p27*x39 + p34*x40 + p41*x41 + p48*x42; 699 pc[41] = m42 = p7*x36 + p14*x37 + p21*x38 + p28*x39 + p35*x40 + p42*x41 + p49*x42; 700 701 pc[42] = m43 = p1*x43 + p8*x44 + p15*x45 + p22*x46 + p29*x47 + p36*x48 + p43*x49; 702 pc[43] = m44 = p2*x43 + p9*x44 + p16*x45 + p23*x46 + p30*x47 + p37*x48 + p44*x49; 703 pc[44] = m45 = p3*x43 + p10*x44 + p17*x45 + p24*x46 + p31*x47 + p38*x48 + p45*x49; 704 pc[45] = m46 = p4*x43 + p11*x44 + p18*x45 + p25*x46 + p32*x47 + p39*x48 + p46*x49; 705 pc[46] = m47 = p5*x43 + p12*x44 + p19*x45 + p26*x46 + p33*x47 + p40*x48 + p47*x49; 706 pc[47] = m48 = p6*x43 + p13*x44 + p20*x45 + p27*x46 + p34*x47 + p41*x48 + p48*x49; 707 pc[48] = m49 = p7*x43 + p14*x44 + p21*x45 + p28*x46 + p35*x47 + p42*x48 + p49*x49; 708 709 nz = bi[row+1] - diag_offset[row] - 1; 710 pv += 49; 711 for (j=0; j<nz; j++) { 712 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 713 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; 714 x9 = pv[8]; x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; 715 x13 = pv[12]; x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; 716 x17 = pv[16]; x18 = pv[17]; x19 = pv[18]; x20 = pv[19]; 717 x21 = pv[20]; x22 = pv[21]; x23 = pv[22]; x24 = pv[23]; 718 x25 = pv[24]; x26 = pv[25]; x27 = pv[26]; x28 = pv[27]; 719 x29 = pv[28]; x30 = pv[29]; x31 = pv[30]; x32 = pv[31]; 720 x33 = pv[32]; x34 = pv[33]; x35 = pv[34]; x36 = pv[35]; 721 x37 = pv[36]; x38 = pv[37]; x39 = pv[38]; x40 = pv[39]; 722 x41 = pv[40]; x42 = pv[41]; x43 = pv[42]; x44 = pv[43]; 723 x45 = pv[44]; x46 = pv[45]; x47 = pv[46]; x48 = pv[47]; 724 x49 = pv[48]; 725 x = rtmp + 49*pj[j]; 726 x[0] -= m1*x1 + m8*x2 + m15*x3 + m22*x4 + m29*x5 + m36*x6 + m43*x7; 727 x[1] -= m2*x1 + m9*x2 + m16*x3 + m23*x4 + m30*x5 + m37*x6 + m44*x7; 728 x[2] -= m3*x1 + m10*x2 + m17*x3 + m24*x4 + m31*x5 + m38*x6 + m45*x7; 729 x[3] -= m4*x1 + m11*x2 + m18*x3 + m25*x4 + m32*x5 + m39*x6 + m46*x7; 730 x[4] -= m5*x1 + m12*x2 + m19*x3 + m26*x4 + m33*x5 + m40*x6 + m47*x7; 731 x[5] -= m6*x1 + m13*x2 + m20*x3 + m27*x4 + m34*x5 + m41*x6 + m48*x7; 732 x[6] -= m7*x1 + m14*x2 + m21*x3 + m28*x4 + m35*x5 + m42*x6 + m49*x7; 733 734 x[7] -= m1*x8 + m8*x9 + m15*x10 + m22*x11 + m29*x12 + m36*x13 + m43*x14; 735 x[8] -= m2*x8 + m9*x9 + m16*x10 + m23*x11 + m30*x12 + m37*x13 + m44*x14; 736 x[9] -= m3*x8 + m10*x9 + m17*x10 + m24*x11 + m31*x12 + m38*x13 + m45*x14; 737 x[10] -= m4*x8 + m11*x9 + m18*x10 + m25*x11 + m32*x12 + m39*x13 + m46*x14; 738 x[11] -= m5*x8 + m12*x9 + m19*x10 + m26*x11 + m33*x12 + m40*x13 + m47*x14; 739 x[12] -= m6*x8 + m13*x9 + m20*x10 + m27*x11 + m34*x12 + m41*x13 + m48*x14; 740 x[13] -= m7*x8 + m14*x9 + m21*x10 + m28*x11 + m35*x12 + m42*x13 + m49*x14; 741 742 x[14] -= m1*x15 + m8*x16 + m15*x17 + m22*x18 + m29*x19 + m36*x20 + m43*x21; 743 x[15] -= m2*x15 + m9*x16 + m16*x17 + m23*x18 + m30*x19 + m37*x20 + m44*x21; 744 x[16] -= m3*x15 + m10*x16 + m17*x17 + m24*x18 + m31*x19 + m38*x20 + m45*x21; 745 x[17] -= m4*x15 + m11*x16 + m18*x17 + m25*x18 + m32*x19 + m39*x20 + m46*x21; 746 x[18] -= m5*x15 + m12*x16 + m19*x17 + m26*x18 + m33*x19 + m40*x20 + m47*x21; 747 x[19] -= m6*x15 + m13*x16 + m20*x17 + m27*x18 + m34*x19 + m41*x20 + m48*x21; 748 x[20] -= m7*x15 + m14*x16 + m21*x17 + m28*x18 + m35*x19 + m42*x20 + m49*x21; 749 750 x[21] -= m1*x22 + m8*x23 + m15*x24 + m22*x25 + m29*x26 + m36*x27 + m43*x28; 751 x[22] -= m2*x22 + m9*x23 + m16*x24 + m23*x25 + m30*x26 + m37*x27 + m44*x28; 752 x[23] -= m3*x22 + m10*x23 + m17*x24 + m24*x25 + m31*x26 + m38*x27 + m45*x28; 753 x[24] -= m4*x22 + m11*x23 + m18*x24 + m25*x25 + m32*x26 + m39*x27 + m46*x28; 754 x[25] -= m5*x22 + m12*x23 + m19*x24 + m26*x25 + m33*x26 + m40*x27 + m47*x28; 755 x[26] -= m6*x22 + m13*x23 + m20*x24 + m27*x25 + m34*x26 + m41*x27 + m48*x28; 756 x[27] -= m7*x22 + m14*x23 + m21*x24 + m28*x25 + m35*x26 + m42*x27 + m49*x28; 757 758 x[28] -= m1*x29 + m8*x30 + m15*x31 + m22*x32 + m29*x33 + m36*x34 + m43*x35; 759 x[29] -= m2*x29 + m9*x30 + m16*x31 + m23*x32 + m30*x33 + m37*x34 + m44*x35; 760 x[30] -= m3*x29 + m10*x30 + m17*x31 + m24*x32 + m31*x33 + m38*x34 + m45*x35; 761 x[31] -= m4*x29 + m11*x30 + m18*x31 + m25*x32 + m32*x33 + m39*x34 + m46*x35; 762 x[32] -= m5*x29 + m12*x30 + m19*x31 + m26*x32 + m33*x33 + m40*x34 + m47*x35; 763 x[33] -= m6*x29 + m13*x30 + m20*x31 + m27*x32 + m34*x33 + m41*x34 + m48*x35; 764 x[34] -= m7*x29 + m14*x30 + m21*x31 + m28*x32 + m35*x33 + m42*x34 + m49*x35; 765 766 x[35] -= m1*x36 + m8*x37 + m15*x38 + m22*x39 + m29*x40 + m36*x41 + m43*x42; 767 x[36] -= m2*x36 + m9*x37 + m16*x38 + m23*x39 + m30*x40 + m37*x41 + m44*x42; 768 x[37] -= m3*x36 + m10*x37 + m17*x38 + m24*x39 + m31*x40 + m38*x41 + m45*x42; 769 x[38] -= m4*x36 + m11*x37 + m18*x38 + m25*x39 + m32*x40 + m39*x41 + m46*x42; 770 x[39] -= m5*x36 + m12*x37 + m19*x38 + m26*x39 + m33*x40 + m40*x41 + m47*x42; 771 x[40] -= m6*x36 + m13*x37 + m20*x38 + m27*x39 + m34*x40 + m41*x41 + m48*x42; 772 x[41] -= m7*x36 + m14*x37 + m21*x38 + m28*x39 + m35*x40 + m42*x41 + m49*x42; 773 774 x[42] -= m1*x43 + m8*x44 + m15*x45 + m22*x46 + m29*x47 + m36*x48 + m43*x49; 775 x[43] -= m2*x43 + m9*x44 + m16*x45 + m23*x46 + m30*x47 + m37*x48 + m44*x49; 776 x[44] -= m3*x43 + m10*x44 + m17*x45 + m24*x46 + m31*x47 + m38*x48 + m45*x49; 777 x[45] -= m4*x43 + m11*x44 + m18*x45 + m25*x46 + m32*x47 + m39*x48 + m46*x49; 778 x[46] -= m5*x43 + m12*x44 + m19*x45 + m26*x46 + m33*x47 + m40*x48 + m47*x49; 779 x[47] -= m6*x43 + m13*x44 + m20*x45 + m27*x46 + m34*x47 + m41*x48 + m48*x49; 780 x[48] -= m7*x43 + m14*x44 + m21*x45 + m28*x46 + m35*x47 + m42*x48 + m49*x49; 781 pv += 49; 782 } 783 PLogFlops(686*nz+637); 784 } 785 row = *ajtmp++; 786 } 787 /* finished row so stick it into b->a */ 788 pv = ba + 49*bi[i]; 789 pj = bj + bi[i]; 790 nz = bi[i+1] - bi[i]; 791 for (j=0; j<nz; j++) { 792 x = rtmp+49*pj[j]; 793 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 794 pv[4] = x[4]; pv[5] = x[5]; pv[6] = x[6]; pv[7] = x[7]; 795 pv[8] = x[8]; pv[9] = x[9]; pv[10] = x[10]; pv[11] = x[11]; 796 pv[12] = x[12]; pv[13] = x[13]; pv[14] = x[14]; pv[15] = x[15]; 797 pv[16] = x[16]; pv[17] = x[17]; pv[18] = x[18]; pv[19] = x[19]; 798 pv[20] = x[20]; pv[21] = x[21]; pv[22] = x[22]; pv[23] = x[23]; 799 pv[24] = x[24]; pv[25] = x[25]; pv[26] = x[26]; pv[27] = x[27]; 800 pv[28] = x[28]; pv[29] = x[29]; pv[30] = x[30]; pv[31] = x[31]; 801 pv[32] = x[32]; pv[33] = x[33]; pv[34] = x[34]; pv[35] = x[35]; 802 pv[36] = x[36]; pv[37] = x[37]; pv[38] = x[38]; pv[39] = x[39]; 803 pv[40] = x[40]; pv[41] = x[41]; pv[42] = x[42]; pv[43] = x[43]; 804 pv[44] = x[44]; pv[45] = x[45]; pv[46] = x[46]; pv[47] = x[47]; 805 pv[48] = x[48]; 806 pv += 49; 807 } 808 /* invert diagonal block */ 809 w = ba + 49*diag_offset[i]; 810 ierr = Kernel_A_gets_inverse_A_7(w);CHKERRQ(ierr); 811 } 812 813 ierr = PetscFree(rtmp);CHKERRQ(ierr); 814 C->factor = FACTOR_LU; 815 C->assembled = PETSC_TRUE; 816 PLogFlops(1.3333*343*b->mbs); /* from inverting diagonal blocks */ 817 PetscFunctionReturn(0); 818 } 819 820 /* ------------------------------------------------------------*/ 821 /* 822 Version for when blocks are 6 by 6 823 */ 824 #undef __FUNC__ 825 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_6" 826 int MatLUFactorNumeric_SeqSBAIJ_6(Mat A,Mat *B) 827 { 828 Mat C = *B; 829 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 830 IS isrow = b->row,isicol = b->icol; 831 int *r,*ic,ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 832 int *ajtmpold,*ajtmp,nz,row; 833 int *diag_offset = b->diag,idx,*ai=a->i,*aj=a->j,*pj; 834 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 835 MatScalar p1,p2,p3,p4,m1,m2,m3,m4,m5,m6,m7,m8,m9,x1,x2,x3,x4; 836 MatScalar p5,p6,p7,p8,p9,x5,x6,x7,x8,x9,x10,x11,x12,x13,x14,x15,x16; 837 MatScalar x17,x18,x19,x20,x21,x22,x23,x24,x25,p10,p11,p12,p13,p14; 838 MatScalar p15,p16,p17,p18,p19,p20,p21,p22,p23,p24,p25,m10,m11,m12; 839 MatScalar m13,m14,m15,m16,m17,m18,m19,m20,m21,m22,m23,m24,m25; 840 MatScalar p26,p27,p28,p29,p30,p31,p32,p33,p34,p35,p36; 841 MatScalar x26,x27,x28,x29,x30,x31,x32,x33,x34,x35,x36; 842 MatScalar m26,m27,m28,m29,m30,m31,m32,m33,m34,m35,m36; 843 MatScalar *ba = b->a,*aa = a->a; 844 845 PetscFunctionBegin; 846 ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr); 847 ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr); 848 rtmp = (MatScalar*)PetscMalloc(36*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 849 850 for (i=0; i<n; i++) { 851 nz = bi[i+1] - bi[i]; 852 ajtmp = bj + bi[i]; 853 for (j=0; j<nz; j++) { 854 x = rtmp+36*ajtmp[j]; 855 x[0] = x[1] = x[2] = x[3] = x[4] = x[5] = x[6] = x[7] = x[8] = x[9] = 0.0; 856 x[10] = x[11] = x[12] = x[13] = x[14] = x[15] = x[16] = x[17] = 0.0; 857 x[18] = x[19] = x[20] = x[21] = x[22] = x[23] = x[24] = x[25] = 0.0 ; 858 x[26] = x[27] = x[28] = x[29] = x[30] = x[31] = x[32] = x[33] = 0.0 ; 859 x[34] = x[35] = 0.0 ; 860 } 861 /* load in initial (unfactored row) */ 862 idx = r[i]; 863 nz = ai[idx+1] - ai[idx]; 864 ajtmpold = aj + ai[idx]; 865 v = aa + 36*ai[idx]; 866 for (j=0; j<nz; j++) { 867 x = rtmp+36*ic[ajtmpold[j]]; 868 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 869 x[4] = v[4]; x[5] = v[5]; x[6] = v[6]; x[7] = v[7]; 870 x[8] = v[8]; x[9] = v[9]; x[10] = v[10]; x[11] = v[11]; 871 x[12] = v[12]; x[13] = v[13]; x[14] = v[14]; x[15] = v[15]; 872 x[16] = v[16]; x[17] = v[17]; x[18] = v[18]; x[19] = v[19]; 873 x[20] = v[20]; x[21] = v[21]; x[22] = v[22]; x[23] = v[23]; 874 x[24] = v[24]; x[25] = v[25]; x[26] = v[26]; x[27] = v[27]; 875 x[28] = v[28]; x[29] = v[29]; x[30] = v[30]; x[31] = v[31]; 876 x[32] = v[32]; x[33] = v[33]; x[34] = v[34]; x[35] = v[35]; 877 v += 36; 878 } 879 row = *ajtmp++; 880 while (row < i) { 881 pc = rtmp + 36*row; 882 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 883 p5 = pc[4]; p6 = pc[5]; p7 = pc[6]; p8 = pc[7]; 884 p9 = pc[8]; p10 = pc[9]; p11 = pc[10]; p12 = pc[11]; 885 p13 = pc[12]; p14 = pc[13]; p15 = pc[14]; p16 = pc[15]; 886 p17 = pc[16]; p18 = pc[17]; p19 = pc[18]; p20 = pc[19]; 887 p21 = pc[20]; p22 = pc[21]; p23 = pc[22]; p24 = pc[23]; 888 p25 = pc[24]; p26 = pc[25]; p27 = pc[26]; p28 = pc[27]; 889 p29 = pc[28]; p30 = pc[29]; p31 = pc[30]; p32 = pc[31]; 890 p33 = pc[32]; p34 = pc[33]; p35 = pc[34]; p36 = pc[35]; 891 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0 || 892 p5 != 0.0 || p6 != 0.0 || p7 != 0.0 || p8 != 0.0 || 893 p9 != 0.0 || p10 != 0.0 || p11 != 0.0 || p12 != 0.0 || 894 p13 != 0.0 || p14 != 0.0 || p15 != 0.0 || p16 != 0.0 || 895 p17 != 0.0 || p18 != 0.0 || p19 != 0.0 || p20 != 0.0 || 896 p21 != 0.0 || p22 != 0.0 || p23 != 0.0 || p24 != 0.0 || 897 p25 != 0.0 || p26 != 0.0 || p27 != 0.0 || p28 != 0.0 || 898 p29 != 0.0 || p30 != 0.0 || p31 != 0.0 || p32 != 0.0 || 899 p33 != 0.0 || p34 != 0.0 || p35 != 0.0 || p36 != 0.0) { 900 pv = ba + 36*diag_offset[row]; 901 pj = bj + diag_offset[row] + 1; 902 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 903 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; 904 x9 = pv[8]; x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; 905 x13 = pv[12]; x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; 906 x17 = pv[16]; x18 = pv[17]; x19 = pv[18]; x20 = pv[19]; 907 x21 = pv[20]; x22 = pv[21]; x23 = pv[22]; x24 = pv[23]; 908 x25 = pv[24]; x26 = pv[25]; x27 = pv[26]; x28 = pv[27]; 909 x29 = pv[28]; x30 = pv[29]; x31 = pv[30]; x32 = pv[31]; 910 x33 = pv[32]; x34 = pv[33]; x35 = pv[34]; x36 = pv[35]; 911 pc[0] = m1 = p1*x1 + p7*x2 + p13*x3 + p19*x4 + p25*x5 + p31*x6; 912 pc[1] = m2 = p2*x1 + p8*x2 + p14*x3 + p20*x4 + p26*x5 + p32*x6; 913 pc[2] = m3 = p3*x1 + p9*x2 + p15*x3 + p21*x4 + p27*x5 + p33*x6; 914 pc[3] = m4 = p4*x1 + p10*x2 + p16*x3 + p22*x4 + p28*x5 + p34*x6; 915 pc[4] = m5 = p5*x1 + p11*x2 + p17*x3 + p23*x4 + p29*x5 + p35*x6; 916 pc[5] = m6 = p6*x1 + p12*x2 + p18*x3 + p24*x4 + p30*x5 + p36*x6; 917 918 pc[6] = m7 = p1*x7 + p7*x8 + p13*x9 + p19*x10 + p25*x11 + p31*x12; 919 pc[7] = m8 = p2*x7 + p8*x8 + p14*x9 + p20*x10 + p26*x11 + p32*x12; 920 pc[8] = m9 = p3*x7 + p9*x8 + p15*x9 + p21*x10 + p27*x11 + p33*x12; 921 pc[9] = m10 = p4*x7 + p10*x8 + p16*x9 + p22*x10 + p28*x11 + p34*x12; 922 pc[10] = m11 = p5*x7 + p11*x8 + p17*x9 + p23*x10 + p29*x11 + p35*x12; 923 pc[11] = m12 = p6*x7 + p12*x8 + p18*x9 + p24*x10 + p30*x11 + p36*x12; 924 925 pc[12] = m13 = p1*x13 + p7*x14 + p13*x15 + p19*x16 + p25*x17 + p31*x18; 926 pc[13] = m14 = p2*x13 + p8*x14 + p14*x15 + p20*x16 + p26*x17 + p32*x18; 927 pc[14] = m15 = p3*x13 + p9*x14 + p15*x15 + p21*x16 + p27*x17 + p33*x18; 928 pc[15] = m16 = p4*x13 + p10*x14 + p16*x15 + p22*x16 + p28*x17 + p34*x18; 929 pc[16] = m17 = p5*x13 + p11*x14 + p17*x15 + p23*x16 + p29*x17 + p35*x18; 930 pc[17] = m18 = p6*x13 + p12*x14 + p18*x15 + p24*x16 + p30*x17 + p36*x18; 931 932 pc[18] = m19 = p1*x19 + p7*x20 + p13*x21 + p19*x22 + p25*x23 + p31*x24; 933 pc[19] = m20 = p2*x19 + p8*x20 + p14*x21 + p20*x22 + p26*x23 + p32*x24; 934 pc[20] = m21 = p3*x19 + p9*x20 + p15*x21 + p21*x22 + p27*x23 + p33*x24; 935 pc[21] = m22 = p4*x19 + p10*x20 + p16*x21 + p22*x22 + p28*x23 + p34*x24; 936 pc[22] = m23 = p5*x19 + p11*x20 + p17*x21 + p23*x22 + p29*x23 + p35*x24; 937 pc[23] = m24 = p6*x19 + p12*x20 + p18*x21 + p24*x22 + p30*x23 + p36*x24; 938 939 pc[24] = m25 = p1*x25 + p7*x26 + p13*x27 + p19*x28 + p25*x29 + p31*x30; 940 pc[25] = m26 = p2*x25 + p8*x26 + p14*x27 + p20*x28 + p26*x29 + p32*x30; 941 pc[26] = m27 = p3*x25 + p9*x26 + p15*x27 + p21*x28 + p27*x29 + p33*x30; 942 pc[27] = m28 = p4*x25 + p10*x26 + p16*x27 + p22*x28 + p28*x29 + p34*x30; 943 pc[28] = m29 = p5*x25 + p11*x26 + p17*x27 + p23*x28 + p29*x29 + p35*x30; 944 pc[29] = m30 = p6*x25 + p12*x26 + p18*x27 + p24*x28 + p30*x29 + p36*x30; 945 946 pc[30] = m31 = p1*x31 + p7*x32 + p13*x33 + p19*x34 + p25*x35 + p31*x36; 947 pc[31] = m32 = p2*x31 + p8*x32 + p14*x33 + p20*x34 + p26*x35 + p32*x36; 948 pc[32] = m33 = p3*x31 + p9*x32 + p15*x33 + p21*x34 + p27*x35 + p33*x36; 949 pc[33] = m34 = p4*x31 + p10*x32 + p16*x33 + p22*x34 + p28*x35 + p34*x36; 950 pc[34] = m35 = p5*x31 + p11*x32 + p17*x33 + p23*x34 + p29*x35 + p35*x36; 951 pc[35] = m36 = p6*x31 + p12*x32 + p18*x33 + p24*x34 + p30*x35 + p36*x36; 952 953 nz = bi[row+1] - diag_offset[row] - 1; 954 pv += 36; 955 for (j=0; j<nz; j++) { 956 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 957 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; 958 x9 = pv[8]; x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; 959 x13 = pv[12]; x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; 960 x17 = pv[16]; x18 = pv[17]; x19 = pv[18]; x20 = pv[19]; 961 x21 = pv[20]; x22 = pv[21]; x23 = pv[22]; x24 = pv[23]; 962 x25 = pv[24]; x26 = pv[25]; x27 = pv[26]; x28 = pv[27]; 963 x29 = pv[28]; x30 = pv[29]; x31 = pv[30]; x32 = pv[31]; 964 x33 = pv[32]; x34 = pv[33]; x35 = pv[34]; x36 = pv[35]; 965 x = rtmp + 36*pj[j]; 966 x[0] -= m1*x1 + m7*x2 + m13*x3 + m19*x4 + m25*x5 + m31*x6; 967 x[1] -= m2*x1 + m8*x2 + m14*x3 + m20*x4 + m26*x5 + m32*x6; 968 x[2] -= m3*x1 + m9*x2 + m15*x3 + m21*x4 + m27*x5 + m33*x6; 969 x[3] -= m4*x1 + m10*x2 + m16*x3 + m22*x4 + m28*x5 + m34*x6; 970 x[4] -= m5*x1 + m11*x2 + m17*x3 + m23*x4 + m29*x5 + m35*x6; 971 x[5] -= m6*x1 + m12*x2 + m18*x3 + m24*x4 + m30*x5 + m36*x6; 972 973 x[6] -= m1*x7 + m7*x8 + m13*x9 + m19*x10 + m25*x11 + m31*x12; 974 x[7] -= m2*x7 + m8*x8 + m14*x9 + m20*x10 + m26*x11 + m32*x12; 975 x[8] -= m3*x7 + m9*x8 + m15*x9 + m21*x10 + m27*x11 + m33*x12; 976 x[9] -= m4*x7 + m10*x8 + m16*x9 + m22*x10 + m28*x11 + m34*x12; 977 x[10] -= m5*x7 + m11*x8 + m17*x9 + m23*x10 + m29*x11 + m35*x12; 978 x[11] -= m6*x7 + m12*x8 + m18*x9 + m24*x10 + m30*x11 + m36*x12; 979 980 x[12] -= m1*x13 + m7*x14 + m13*x15 + m19*x16 + m25*x17 + m31*x18; 981 x[13] -= m2*x13 + m8*x14 + m14*x15 + m20*x16 + m26*x17 + m32*x18; 982 x[14] -= m3*x13 + m9*x14 + m15*x15 + m21*x16 + m27*x17 + m33*x18; 983 x[15] -= m4*x13 + m10*x14 + m16*x15 + m22*x16 + m28*x17 + m34*x18; 984 x[16] -= m5*x13 + m11*x14 + m17*x15 + m23*x16 + m29*x17 + m35*x18; 985 x[17] -= m6*x13 + m12*x14 + m18*x15 + m24*x16 + m30*x17 + m36*x18; 986 987 x[18] -= m1*x19 + m7*x20 + m13*x21 + m19*x22 + m25*x23 + m31*x24; 988 x[19] -= m2*x19 + m8*x20 + m14*x21 + m20*x22 + m26*x23 + m32*x24; 989 x[20] -= m3*x19 + m9*x20 + m15*x21 + m21*x22 + m27*x23 + m33*x24; 990 x[21] -= m4*x19 + m10*x20 + m16*x21 + m22*x22 + m28*x23 + m34*x24; 991 x[22] -= m5*x19 + m11*x20 + m17*x21 + m23*x22 + m29*x23 + m35*x24; 992 x[23] -= m6*x19 + m12*x20 + m18*x21 + m24*x22 + m30*x23 + m36*x24; 993 994 x[24] -= m1*x25 + m7*x26 + m13*x27 + m19*x28 + m25*x29 + m31*x30; 995 x[25] -= m2*x25 + m8*x26 + m14*x27 + m20*x28 + m26*x29 + m32*x30; 996 x[26] -= m3*x25 + m9*x26 + m15*x27 + m21*x28 + m27*x29 + m33*x30; 997 x[27] -= m4*x25 + m10*x26 + m16*x27 + m22*x28 + m28*x29 + m34*x30; 998 x[28] -= m5*x25 + m11*x26 + m17*x27 + m23*x28 + m29*x29 + m35*x30; 999 x[29] -= m6*x25 + m12*x26 + m18*x27 + m24*x28 + m30*x29 + m36*x30; 1000 1001 x[30] -= m1*x31 + m7*x32 + m13*x33 + m19*x34 + m25*x35 + m31*x36; 1002 x[31] -= m2*x31 + m8*x32 + m14*x33 + m20*x34 + m26*x35 + m32*x36; 1003 x[32] -= m3*x31 + m9*x32 + m15*x33 + m21*x34 + m27*x35 + m33*x36; 1004 x[33] -= m4*x31 + m10*x32 + m16*x33 + m22*x34 + m28*x35 + m34*x36; 1005 x[34] -= m5*x31 + m11*x32 + m17*x33 + m23*x34 + m29*x35 + m35*x36; 1006 x[35] -= m6*x31 + m12*x32 + m18*x33 + m24*x34 + m30*x35 + m36*x36; 1007 1008 pv += 36; 1009 } 1010 PLogFlops(432*nz+396); 1011 } 1012 row = *ajtmp++; 1013 } 1014 /* finished row so stick it into b->a */ 1015 pv = ba + 36*bi[i]; 1016 pj = bj + bi[i]; 1017 nz = bi[i+1] - bi[i]; 1018 for (j=0; j<nz; j++) { 1019 x = rtmp+36*pj[j]; 1020 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 1021 pv[4] = x[4]; pv[5] = x[5]; pv[6] = x[6]; pv[7] = x[7]; 1022 pv[8] = x[8]; pv[9] = x[9]; pv[10] = x[10]; pv[11] = x[11]; 1023 pv[12] = x[12]; pv[13] = x[13]; pv[14] = x[14]; pv[15] = x[15]; 1024 pv[16] = x[16]; pv[17] = x[17]; pv[18] = x[18]; pv[19] = x[19]; 1025 pv[20] = x[20]; pv[21] = x[21]; pv[22] = x[22]; pv[23] = x[23]; 1026 pv[24] = x[24]; pv[25] = x[25]; pv[26] = x[26]; pv[27] = x[27]; 1027 pv[28] = x[28]; pv[29] = x[29]; pv[30] = x[30]; pv[31] = x[31]; 1028 pv[32] = x[32]; pv[33] = x[33]; pv[34] = x[34]; pv[35] = x[35]; 1029 pv += 36; 1030 } 1031 /* invert diagonal block */ 1032 w = ba + 36*diag_offset[i]; 1033 ierr = Kernel_A_gets_inverse_A_6(w);CHKERRQ(ierr); 1034 } 1035 1036 ierr = PetscFree(rtmp);CHKERRQ(ierr); 1037 ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr); 1038 ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr); 1039 C->factor = FACTOR_LU; 1040 C->assembled = PETSC_TRUE; 1041 PLogFlops(1.3333*216*b->mbs); /* from inverting diagonal blocks */ 1042 PetscFunctionReturn(0); 1043 } 1044 /* 1045 Version for when blocks are 6 by 6 Using natural ordering 1046 */ 1047 #undef __FUNC__ 1048 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_6_NaturalOrdering" 1049 int MatLUFactorNumeric_SeqSBAIJ_6_NaturalOrdering(Mat A,Mat *B) 1050 { 1051 Mat C = *B; 1052 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 1053 int ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 1054 int *ajtmpold,*ajtmp,nz,row; 1055 int *diag_offset = b->diag,*ai=a->i,*aj=a->j,*pj; 1056 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 1057 MatScalar x1,x2,x3,x4,x5,x6,x7,x8,x9,x10,x11,x12,x13,x14,x15; 1058 MatScalar x16,x17,x18,x19,x20,x21,x22,x23,x24,x25; 1059 MatScalar p1,p2,p3,p4,p5,p6,p7,p8,p9,p10,p11,p12,p13,p14,p15; 1060 MatScalar p16,p17,p18,p19,p20,p21,p22,p23,p24,p25; 1061 MatScalar m1,m2,m3,m4,m5,m6,m7,m8,m9,m10,m11,m12,m13,m14,m15; 1062 MatScalar m16,m17,m18,m19,m20,m21,m22,m23,m24,m25; 1063 MatScalar p26,p27,p28,p29,p30,p31,p32,p33,p34,p35,p36; 1064 MatScalar x26,x27,x28,x29,x30,x31,x32,x33,x34,x35,x36; 1065 MatScalar m26,m27,m28,m29,m30,m31,m32,m33,m34,m35,m36; 1066 MatScalar *ba = b->a,*aa = a->a; 1067 1068 PetscFunctionBegin; 1069 rtmp = (MatScalar*)PetscMalloc(36*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 1070 for (i=0; i<n; i++) { 1071 nz = bi[i+1] - bi[i]; 1072 ajtmp = bj + bi[i]; 1073 for (j=0; j<nz; j++) { 1074 x = rtmp+36*ajtmp[j]; 1075 x[0] = x[1] = x[2] = x[3] = x[4] = x[5] = x[6] = x[7] = x[8] = x[9] = 0.0; 1076 x[10] = x[11] = x[12] = x[13] = x[14] = x[15] = x[16] = x[17] = 0.0; 1077 x[18] = x[19] = x[20] = x[21] = x[22] = x[23] = x[24] = x[25] = 0.0 ; 1078 x[26] = x[27] = x[28] = x[29] = x[30] = x[31] = x[32] = x[33] = 0.0 ; 1079 x[34] = x[35] = 0.0 ; 1080 } 1081 /* load in initial (unfactored row) */ 1082 nz = ai[i+1] - ai[i]; 1083 ajtmpold = aj + ai[i]; 1084 v = aa + 36*ai[i]; 1085 for (j=0; j<nz; j++) { 1086 x = rtmp+36*ajtmpold[j]; 1087 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 1088 x[4] = v[4]; x[5] = v[5]; x[6] = v[6]; x[7] = v[7]; 1089 x[8] = v[8]; x[9] = v[9]; x[10] = v[10]; x[11] = v[11]; 1090 x[12] = v[12]; x[13] = v[13]; x[14] = v[14]; x[15] = v[15]; 1091 x[16] = v[16]; x[17] = v[17]; x[18] = v[18]; x[19] = v[19]; 1092 x[20] = v[20]; x[21] = v[21]; x[22] = v[22]; x[23] = v[23]; 1093 x[24] = v[24]; x[25] = v[25]; x[26] = v[26]; x[27] = v[27]; 1094 x[28] = v[28]; x[29] = v[29]; x[30] = v[30]; x[31] = v[31]; 1095 x[32] = v[32]; x[33] = v[33]; x[34] = v[34]; x[35] = v[35]; 1096 v += 36; 1097 } 1098 row = *ajtmp++; 1099 while (row < i) { 1100 pc = rtmp + 36*row; 1101 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 1102 p5 = pc[4]; p6 = pc[5]; p7 = pc[6]; p8 = pc[7]; 1103 p9 = pc[8]; p10 = pc[9]; p11 = pc[10]; p12 = pc[11]; 1104 p13 = pc[12]; p14 = pc[13]; p15 = pc[14]; p16 = pc[15]; 1105 p17 = pc[16]; p18 = pc[17]; p19 = pc[18]; p20 = pc[19]; 1106 p21 = pc[20]; p22 = pc[21]; p23 = pc[22]; p24 = pc[23]; 1107 p25 = pc[24]; p26 = pc[25]; p27 = pc[26]; p28 = pc[27]; 1108 p29 = pc[28]; p30 = pc[29]; p31 = pc[30]; p32 = pc[31]; 1109 p33 = pc[32]; p34 = pc[33]; p35 = pc[34]; p36 = pc[35]; 1110 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0 || 1111 p5 != 0.0 || p6 != 0.0 || p7 != 0.0 || p8 != 0.0 || 1112 p9 != 0.0 || p10 != 0.0 || p11 != 0.0 || p12 != 0.0 || 1113 p13 != 0.0 || p14 != 0.0 || p15 != 0.0 || p16 != 0.0 || 1114 p17 != 0.0 || p18 != 0.0 || p19 != 0.0 || p20 != 0.0 || 1115 p21 != 0.0 || p22 != 0.0 || p23 != 0.0 || p24 != 0.0 || 1116 p25 != 0.0 || p26 != 0.0 || p27 != 0.0 || p28 != 0.0 || 1117 p29 != 0.0 || p30 != 0.0 || p31 != 0.0 || p32 != 0.0 || 1118 p33 != 0.0 || p34 != 0.0 || p35 != 0.0 || p36 != 0.0) { 1119 pv = ba + 36*diag_offset[row]; 1120 pj = bj + diag_offset[row] + 1; 1121 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1122 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; 1123 x9 = pv[8]; x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; 1124 x13 = pv[12]; x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; 1125 x17 = pv[16]; x18 = pv[17]; x19 = pv[18]; x20 = pv[19]; 1126 x21 = pv[20]; x22 = pv[21]; x23 = pv[22]; x24 = pv[23]; 1127 x25 = pv[24]; x26 = pv[25]; x27 = pv[26]; x28 = pv[27]; 1128 x29 = pv[28]; x30 = pv[29]; x31 = pv[30]; x32 = pv[31]; 1129 x33 = pv[32]; x34 = pv[33]; x35 = pv[34]; x36 = pv[35]; 1130 pc[0] = m1 = p1*x1 + p7*x2 + p13*x3 + p19*x4 + p25*x5 + p31*x6; 1131 pc[1] = m2 = p2*x1 + p8*x2 + p14*x3 + p20*x4 + p26*x5 + p32*x6; 1132 pc[2] = m3 = p3*x1 + p9*x2 + p15*x3 + p21*x4 + p27*x5 + p33*x6; 1133 pc[3] = m4 = p4*x1 + p10*x2 + p16*x3 + p22*x4 + p28*x5 + p34*x6; 1134 pc[4] = m5 = p5*x1 + p11*x2 + p17*x3 + p23*x4 + p29*x5 + p35*x6; 1135 pc[5] = m6 = p6*x1 + p12*x2 + p18*x3 + p24*x4 + p30*x5 + p36*x6; 1136 1137 pc[6] = m7 = p1*x7 + p7*x8 + p13*x9 + p19*x10 + p25*x11 + p31*x12; 1138 pc[7] = m8 = p2*x7 + p8*x8 + p14*x9 + p20*x10 + p26*x11 + p32*x12; 1139 pc[8] = m9 = p3*x7 + p9*x8 + p15*x9 + p21*x10 + p27*x11 + p33*x12; 1140 pc[9] = m10 = p4*x7 + p10*x8 + p16*x9 + p22*x10 + p28*x11 + p34*x12; 1141 pc[10] = m11 = p5*x7 + p11*x8 + p17*x9 + p23*x10 + p29*x11 + p35*x12; 1142 pc[11] = m12 = p6*x7 + p12*x8 + p18*x9 + p24*x10 + p30*x11 + p36*x12; 1143 1144 pc[12] = m13 = p1*x13 + p7*x14 + p13*x15 + p19*x16 + p25*x17 + p31*x18; 1145 pc[13] = m14 = p2*x13 + p8*x14 + p14*x15 + p20*x16 + p26*x17 + p32*x18; 1146 pc[14] = m15 = p3*x13 + p9*x14 + p15*x15 + p21*x16 + p27*x17 + p33*x18; 1147 pc[15] = m16 = p4*x13 + p10*x14 + p16*x15 + p22*x16 + p28*x17 + p34*x18; 1148 pc[16] = m17 = p5*x13 + p11*x14 + p17*x15 + p23*x16 + p29*x17 + p35*x18; 1149 pc[17] = m18 = p6*x13 + p12*x14 + p18*x15 + p24*x16 + p30*x17 + p36*x18; 1150 1151 pc[18] = m19 = p1*x19 + p7*x20 + p13*x21 + p19*x22 + p25*x23 + p31*x24; 1152 pc[19] = m20 = p2*x19 + p8*x20 + p14*x21 + p20*x22 + p26*x23 + p32*x24; 1153 pc[20] = m21 = p3*x19 + p9*x20 + p15*x21 + p21*x22 + p27*x23 + p33*x24; 1154 pc[21] = m22 = p4*x19 + p10*x20 + p16*x21 + p22*x22 + p28*x23 + p34*x24; 1155 pc[22] = m23 = p5*x19 + p11*x20 + p17*x21 + p23*x22 + p29*x23 + p35*x24; 1156 pc[23] = m24 = p6*x19 + p12*x20 + p18*x21 + p24*x22 + p30*x23 + p36*x24; 1157 1158 pc[24] = m25 = p1*x25 + p7*x26 + p13*x27 + p19*x28 + p25*x29 + p31*x30; 1159 pc[25] = m26 = p2*x25 + p8*x26 + p14*x27 + p20*x28 + p26*x29 + p32*x30; 1160 pc[26] = m27 = p3*x25 + p9*x26 + p15*x27 + p21*x28 + p27*x29 + p33*x30; 1161 pc[27] = m28 = p4*x25 + p10*x26 + p16*x27 + p22*x28 + p28*x29 + p34*x30; 1162 pc[28] = m29 = p5*x25 + p11*x26 + p17*x27 + p23*x28 + p29*x29 + p35*x30; 1163 pc[29] = m30 = p6*x25 + p12*x26 + p18*x27 + p24*x28 + p30*x29 + p36*x30; 1164 1165 pc[30] = m31 = p1*x31 + p7*x32 + p13*x33 + p19*x34 + p25*x35 + p31*x36; 1166 pc[31] = m32 = p2*x31 + p8*x32 + p14*x33 + p20*x34 + p26*x35 + p32*x36; 1167 pc[32] = m33 = p3*x31 + p9*x32 + p15*x33 + p21*x34 + p27*x35 + p33*x36; 1168 pc[33] = m34 = p4*x31 + p10*x32 + p16*x33 + p22*x34 + p28*x35 + p34*x36; 1169 pc[34] = m35 = p5*x31 + p11*x32 + p17*x33 + p23*x34 + p29*x35 + p35*x36; 1170 pc[35] = m36 = p6*x31 + p12*x32 + p18*x33 + p24*x34 + p30*x35 + p36*x36; 1171 1172 nz = bi[row+1] - diag_offset[row] - 1; 1173 pv += 36; 1174 for (j=0; j<nz; j++) { 1175 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1176 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; 1177 x9 = pv[8]; x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; 1178 x13 = pv[12]; x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; 1179 x17 = pv[16]; x18 = pv[17]; x19 = pv[18]; x20 = pv[19]; 1180 x21 = pv[20]; x22 = pv[21]; x23 = pv[22]; x24 = pv[23]; 1181 x25 = pv[24]; x26 = pv[25]; x27 = pv[26]; x28 = pv[27]; 1182 x29 = pv[28]; x30 = pv[29]; x31 = pv[30]; x32 = pv[31]; 1183 x33 = pv[32]; x34 = pv[33]; x35 = pv[34]; x36 = pv[35]; 1184 x = rtmp + 36*pj[j]; 1185 x[0] -= m1*x1 + m7*x2 + m13*x3 + m19*x4 + m25*x5 + m31*x6; 1186 x[1] -= m2*x1 + m8*x2 + m14*x3 + m20*x4 + m26*x5 + m32*x6; 1187 x[2] -= m3*x1 + m9*x2 + m15*x3 + m21*x4 + m27*x5 + m33*x6; 1188 x[3] -= m4*x1 + m10*x2 + m16*x3 + m22*x4 + m28*x5 + m34*x6; 1189 x[4] -= m5*x1 + m11*x2 + m17*x3 + m23*x4 + m29*x5 + m35*x6; 1190 x[5] -= m6*x1 + m12*x2 + m18*x3 + m24*x4 + m30*x5 + m36*x6; 1191 1192 x[6] -= m1*x7 + m7*x8 + m13*x9 + m19*x10 + m25*x11 + m31*x12; 1193 x[7] -= m2*x7 + m8*x8 + m14*x9 + m20*x10 + m26*x11 + m32*x12; 1194 x[8] -= m3*x7 + m9*x8 + m15*x9 + m21*x10 + m27*x11 + m33*x12; 1195 x[9] -= m4*x7 + m10*x8 + m16*x9 + m22*x10 + m28*x11 + m34*x12; 1196 x[10] -= m5*x7 + m11*x8 + m17*x9 + m23*x10 + m29*x11 + m35*x12; 1197 x[11] -= m6*x7 + m12*x8 + m18*x9 + m24*x10 + m30*x11 + m36*x12; 1198 1199 x[12] -= m1*x13 + m7*x14 + m13*x15 + m19*x16 + m25*x17 + m31*x18; 1200 x[13] -= m2*x13 + m8*x14 + m14*x15 + m20*x16 + m26*x17 + m32*x18; 1201 x[14] -= m3*x13 + m9*x14 + m15*x15 + m21*x16 + m27*x17 + m33*x18; 1202 x[15] -= m4*x13 + m10*x14 + m16*x15 + m22*x16 + m28*x17 + m34*x18; 1203 x[16] -= m5*x13 + m11*x14 + m17*x15 + m23*x16 + m29*x17 + m35*x18; 1204 x[17] -= m6*x13 + m12*x14 + m18*x15 + m24*x16 + m30*x17 + m36*x18; 1205 1206 x[18] -= m1*x19 + m7*x20 + m13*x21 + m19*x22 + m25*x23 + m31*x24; 1207 x[19] -= m2*x19 + m8*x20 + m14*x21 + m20*x22 + m26*x23 + m32*x24; 1208 x[20] -= m3*x19 + m9*x20 + m15*x21 + m21*x22 + m27*x23 + m33*x24; 1209 x[21] -= m4*x19 + m10*x20 + m16*x21 + m22*x22 + m28*x23 + m34*x24; 1210 x[22] -= m5*x19 + m11*x20 + m17*x21 + m23*x22 + m29*x23 + m35*x24; 1211 x[23] -= m6*x19 + m12*x20 + m18*x21 + m24*x22 + m30*x23 + m36*x24; 1212 1213 x[24] -= m1*x25 + m7*x26 + m13*x27 + m19*x28 + m25*x29 + m31*x30; 1214 x[25] -= m2*x25 + m8*x26 + m14*x27 + m20*x28 + m26*x29 + m32*x30; 1215 x[26] -= m3*x25 + m9*x26 + m15*x27 + m21*x28 + m27*x29 + m33*x30; 1216 x[27] -= m4*x25 + m10*x26 + m16*x27 + m22*x28 + m28*x29 + m34*x30; 1217 x[28] -= m5*x25 + m11*x26 + m17*x27 + m23*x28 + m29*x29 + m35*x30; 1218 x[29] -= m6*x25 + m12*x26 + m18*x27 + m24*x28 + m30*x29 + m36*x30; 1219 1220 x[30] -= m1*x31 + m7*x32 + m13*x33 + m19*x34 + m25*x35 + m31*x36; 1221 x[31] -= m2*x31 + m8*x32 + m14*x33 + m20*x34 + m26*x35 + m32*x36; 1222 x[32] -= m3*x31 + m9*x32 + m15*x33 + m21*x34 + m27*x35 + m33*x36; 1223 x[33] -= m4*x31 + m10*x32 + m16*x33 + m22*x34 + m28*x35 + m34*x36; 1224 x[34] -= m5*x31 + m11*x32 + m17*x33 + m23*x34 + m29*x35 + m35*x36; 1225 x[35] -= m6*x31 + m12*x32 + m18*x33 + m24*x34 + m30*x35 + m36*x36; 1226 1227 pv += 36; 1228 } 1229 PLogFlops(432*nz+396); 1230 } 1231 row = *ajtmp++; 1232 } 1233 /* finished row so stick it into b->a */ 1234 pv = ba + 36*bi[i]; 1235 pj = bj + bi[i]; 1236 nz = bi[i+1] - bi[i]; 1237 for (j=0; j<nz; j++) { 1238 x = rtmp+36*pj[j]; 1239 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 1240 pv[4] = x[4]; pv[5] = x[5]; pv[6] = x[6]; pv[7] = x[7]; 1241 pv[8] = x[8]; pv[9] = x[9]; pv[10] = x[10]; pv[11] = x[11]; 1242 pv[12] = x[12]; pv[13] = x[13]; pv[14] = x[14]; pv[15] = x[15]; 1243 pv[16] = x[16]; pv[17] = x[17]; pv[18] = x[18]; pv[19] = x[19]; 1244 pv[20] = x[20]; pv[21] = x[21]; pv[22] = x[22]; pv[23] = x[23]; 1245 pv[24] = x[24]; pv[25] = x[25]; pv[26] = x[26]; pv[27] = x[27]; 1246 pv[28] = x[28]; pv[29] = x[29]; pv[30] = x[30]; pv[31] = x[31]; 1247 pv[32] = x[32]; pv[33] = x[33]; pv[34] = x[34]; pv[35] = x[35]; 1248 pv += 36; 1249 } 1250 /* invert diagonal block */ 1251 w = ba + 36*diag_offset[i]; 1252 ierr = Kernel_A_gets_inverse_A_6(w);CHKERRQ(ierr); 1253 } 1254 1255 ierr = PetscFree(rtmp);CHKERRQ(ierr); 1256 C->factor = FACTOR_LU; 1257 C->assembled = PETSC_TRUE; 1258 PLogFlops(1.3333*216*b->mbs); /* from inverting diagonal blocks */ 1259 PetscFunctionReturn(0); 1260 } 1261 1262 /* ------------------------------------------------------------*/ 1263 /* 1264 Version for when blocks are 5 by 5 1265 */ 1266 #undef __FUNC__ 1267 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_5" 1268 int MatLUFactorNumeric_SeqSBAIJ_5(Mat A,Mat *B) 1269 { 1270 Mat C = *B; 1271 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 1272 IS isrow = b->row,isicol = b->icol; 1273 int *r,*ic,ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 1274 int *ajtmpold,*ajtmp,nz,row; 1275 int *diag_offset = b->diag,idx,*ai=a->i,*aj=a->j,*pj; 1276 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 1277 MatScalar p1,p2,p3,p4,m1,m2,m3,m4,m5,m6,m7,m8,m9,x1,x2,x3,x4; 1278 MatScalar p5,p6,p7,p8,p9,x5,x6,x7,x8,x9,x10,x11,x12,x13,x14,x15,x16; 1279 MatScalar x17,x18,x19,x20,x21,x22,x23,x24,x25,p10,p11,p12,p13,p14; 1280 MatScalar p15,p16,p17,p18,p19,p20,p21,p22,p23,p24,p25,m10,m11,m12; 1281 MatScalar m13,m14,m15,m16,m17,m18,m19,m20,m21,m22,m23,m24,m25; 1282 MatScalar *ba = b->a,*aa = a->a; 1283 1284 PetscFunctionBegin; 1285 ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr); 1286 ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr); 1287 rtmp = (MatScalar*)PetscMalloc(25*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 1288 1289 for (i=0; i<n; i++) { 1290 nz = bi[i+1] - bi[i]; 1291 ajtmp = bj + bi[i]; 1292 for (j=0; j<nz; j++) { 1293 x = rtmp+25*ajtmp[j]; 1294 x[0] = x[1] = x[2] = x[3] = x[4] = x[5] = x[6] = x[7] = x[8] = x[9] = 0.0; 1295 x[10] = x[11] = x[12] = x[13] = x[14] = x[15] = x[16] = x[17] = 0.0; 1296 x[18] = x[19] = x[20] = x[21] = x[22] = x[23] = x[24] = 0.0; 1297 } 1298 /* load in initial (unfactored row) */ 1299 idx = r[i]; 1300 nz = ai[idx+1] - ai[idx]; 1301 ajtmpold = aj + ai[idx]; 1302 v = aa + 25*ai[idx]; 1303 for (j=0; j<nz; j++) { 1304 x = rtmp+25*ic[ajtmpold[j]]; 1305 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 1306 x[4] = v[4]; x[5] = v[5]; x[6] = v[6]; x[7] = v[7]; x[8] = v[8]; 1307 x[9] = v[9]; x[10] = v[10]; x[11] = v[11]; x[12] = v[12]; x[13] = v[13]; 1308 x[14] = v[14]; x[15] = v[15]; x[16] = v[16]; x[17] = v[17]; 1309 x[18] = v[18]; x[19] = v[19]; x[20] = v[20]; x[21] = v[21]; 1310 x[22] = v[22]; x[23] = v[23]; x[24] = v[24]; 1311 v += 25; 1312 } 1313 row = *ajtmp++; 1314 while (row < i) { 1315 pc = rtmp + 25*row; 1316 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 1317 p5 = pc[4]; p6 = pc[5]; p7 = pc[6]; p8 = pc[7]; p9 = pc[8]; 1318 p10 = pc[9]; p11 = pc[10]; p12 = pc[11]; p13 = pc[12]; p14 = pc[13]; 1319 p15 = pc[14]; p16 = pc[15]; p17 = pc[16]; p18 = pc[17]; p19 = pc[18]; 1320 p20 = pc[19]; p21 = pc[20]; p22 = pc[21]; p23 = pc[22]; p24 = pc[23]; 1321 p25 = pc[24]; 1322 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0 || p5 != 0.0 || 1323 p6 != 0.0 || p7 != 0.0 || p8 != 0.0 || p9 != 0.0 || p10 != 0.0 || 1324 p11 != 0.0 || p12 != 0.0 || p13 != 0.0 || p14 != 0.0 || p15 != 0.0 1325 || p16 != 0.0 || p17 != 0.0 || p18 != 0.0 || p19 != 0.0 || 1326 p20 != 0.0 || p21 != 0.0 || p22 != 0.0 || p23 != 0.0 || 1327 p24 != 0.0 || p25 != 0.0) { 1328 pv = ba + 25*diag_offset[row]; 1329 pj = bj + diag_offset[row] + 1; 1330 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1331 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 1332 x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; x13 = pv[12]; x14 = pv[13]; 1333 x15 = pv[14]; x16 = pv[15]; x17 = pv[16]; x18 = pv[17]; 1334 x19 = pv[18]; x20 = pv[19]; x21 = pv[20]; x22 = pv[21]; 1335 x23 = pv[22]; x24 = pv[23]; x25 = pv[24]; 1336 pc[0] = m1 = p1*x1 + p6*x2 + p11*x3 + p16*x4 + p21*x5; 1337 pc[1] = m2 = p2*x1 + p7*x2 + p12*x3 + p17*x4 + p22*x5; 1338 pc[2] = m3 = p3*x1 + p8*x2 + p13*x3 + p18*x4 + p23*x5; 1339 pc[3] = m4 = p4*x1 + p9*x2 + p14*x3 + p19*x4 + p24*x5; 1340 pc[4] = m5 = p5*x1 + p10*x2 + p15*x3 + p20*x4 + p25*x5; 1341 1342 pc[5] = m6 = p1*x6 + p6*x7 + p11*x8 + p16*x9 + p21*x10; 1343 pc[6] = m7 = p2*x6 + p7*x7 + p12*x8 + p17*x9 + p22*x10; 1344 pc[7] = m8 = p3*x6 + p8*x7 + p13*x8 + p18*x9 + p23*x10; 1345 pc[8] = m9 = p4*x6 + p9*x7 + p14*x8 + p19*x9 + p24*x10; 1346 pc[9] = m10 = p5*x6 + p10*x7 + p15*x8 + p20*x9 + p25*x10; 1347 1348 pc[10] = m11 = p1*x11 + p6*x12 + p11*x13 + p16*x14 + p21*x15; 1349 pc[11] = m12 = p2*x11 + p7*x12 + p12*x13 + p17*x14 + p22*x15; 1350 pc[12] = m13 = p3*x11 + p8*x12 + p13*x13 + p18*x14 + p23*x15; 1351 pc[13] = m14 = p4*x11 + p9*x12 + p14*x13 + p19*x14 + p24*x15; 1352 pc[14] = m15 = p5*x11 + p10*x12 + p15*x13 + p20*x14 + p25*x15; 1353 1354 pc[15] = m16 = p1*x16 + p6*x17 + p11*x18 + p16*x19 + p21*x20; 1355 pc[16] = m17 = p2*x16 + p7*x17 + p12*x18 + p17*x19 + p22*x20; 1356 pc[17] = m18 = p3*x16 + p8*x17 + p13*x18 + p18*x19 + p23*x20; 1357 pc[18] = m19 = p4*x16 + p9*x17 + p14*x18 + p19*x19 + p24*x20; 1358 pc[19] = m20 = p5*x16 + p10*x17 + p15*x18 + p20*x19 + p25*x20; 1359 1360 pc[20] = m21 = p1*x21 + p6*x22 + p11*x23 + p16*x24 + p21*x25; 1361 pc[21] = m22 = p2*x21 + p7*x22 + p12*x23 + p17*x24 + p22*x25; 1362 pc[22] = m23 = p3*x21 + p8*x22 + p13*x23 + p18*x24 + p23*x25; 1363 pc[23] = m24 = p4*x21 + p9*x22 + p14*x23 + p19*x24 + p24*x25; 1364 pc[24] = m25 = p5*x21 + p10*x22 + p15*x23 + p20*x24 + p25*x25; 1365 1366 nz = bi[row+1] - diag_offset[row] - 1; 1367 pv += 25; 1368 for (j=0; j<nz; j++) { 1369 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1370 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 1371 x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; x13 = pv[12]; 1372 x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; x17 = pv[16]; 1373 x18 = pv[17]; x19 = pv[18]; x20 = pv[19]; x21 = pv[20]; 1374 x22 = pv[21]; x23 = pv[22]; x24 = pv[23]; x25 = pv[24]; 1375 x = rtmp + 25*pj[j]; 1376 x[0] -= m1*x1 + m6*x2 + m11*x3 + m16*x4 + m21*x5; 1377 x[1] -= m2*x1 + m7*x2 + m12*x3 + m17*x4 + m22*x5; 1378 x[2] -= m3*x1 + m8*x2 + m13*x3 + m18*x4 + m23*x5; 1379 x[3] -= m4*x1 + m9*x2 + m14*x3 + m19*x4 + m24*x5; 1380 x[4] -= m5*x1 + m10*x2 + m15*x3 + m20*x4 + m25*x5; 1381 1382 x[5] -= m1*x6 + m6*x7 + m11*x8 + m16*x9 + m21*x10; 1383 x[6] -= m2*x6 + m7*x7 + m12*x8 + m17*x9 + m22*x10; 1384 x[7] -= m3*x6 + m8*x7 + m13*x8 + m18*x9 + m23*x10; 1385 x[8] -= m4*x6 + m9*x7 + m14*x8 + m19*x9 + m24*x10; 1386 x[9] -= m5*x6 + m10*x7 + m15*x8 + m20*x9 + m25*x10; 1387 1388 x[10] -= m1*x11 + m6*x12 + m11*x13 + m16*x14 + m21*x15; 1389 x[11] -= m2*x11 + m7*x12 + m12*x13 + m17*x14 + m22*x15; 1390 x[12] -= m3*x11 + m8*x12 + m13*x13 + m18*x14 + m23*x15; 1391 x[13] -= m4*x11 + m9*x12 + m14*x13 + m19*x14 + m24*x15; 1392 x[14] -= m5*x11 + m10*x12 + m15*x13 + m20*x14 + m25*x15; 1393 1394 x[15] -= m1*x16 + m6*x17 + m11*x18 + m16*x19 + m21*x20; 1395 x[16] -= m2*x16 + m7*x17 + m12*x18 + m17*x19 + m22*x20; 1396 x[17] -= m3*x16 + m8*x17 + m13*x18 + m18*x19 + m23*x20; 1397 x[18] -= m4*x16 + m9*x17 + m14*x18 + m19*x19 + m24*x20; 1398 x[19] -= m5*x16 + m10*x17 + m15*x18 + m20*x19 + m25*x20; 1399 1400 x[20] -= m1*x21 + m6*x22 + m11*x23 + m16*x24 + m21*x25; 1401 x[21] -= m2*x21 + m7*x22 + m12*x23 + m17*x24 + m22*x25; 1402 x[22] -= m3*x21 + m8*x22 + m13*x23 + m18*x24 + m23*x25; 1403 x[23] -= m4*x21 + m9*x22 + m14*x23 + m19*x24 + m24*x25; 1404 x[24] -= m5*x21 + m10*x22 + m15*x23 + m20*x24 + m25*x25; 1405 1406 pv += 25; 1407 } 1408 PLogFlops(250*nz+225); 1409 } 1410 row = *ajtmp++; 1411 } 1412 /* finished row so stick it into b->a */ 1413 pv = ba + 25*bi[i]; 1414 pj = bj + bi[i]; 1415 nz = bi[i+1] - bi[i]; 1416 for (j=0; j<nz; j++) { 1417 x = rtmp+25*pj[j]; 1418 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 1419 pv[4] = x[4]; pv[5] = x[5]; pv[6] = x[6]; pv[7] = x[7]; pv[8] = x[8]; 1420 pv[9] = x[9]; pv[10] = x[10]; pv[11] = x[11]; pv[12] = x[12]; 1421 pv[13] = x[13]; pv[14] = x[14]; pv[15] = x[15]; pv[16] = x[16]; 1422 pv[17] = x[17]; pv[18] = x[18]; pv[19] = x[19]; pv[20] = x[20]; 1423 pv[21] = x[21]; pv[22] = x[22]; pv[23] = x[23]; pv[24] = x[24]; 1424 pv += 25; 1425 } 1426 /* invert diagonal block */ 1427 w = ba + 25*diag_offset[i]; 1428 ierr = Kernel_A_gets_inverse_A_5(w);CHKERRQ(ierr); 1429 } 1430 1431 ierr = PetscFree(rtmp);CHKERRQ(ierr); 1432 ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr); 1433 ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr); 1434 C->factor = FACTOR_LU; 1435 C->assembled = PETSC_TRUE; 1436 PLogFlops(1.3333*125*b->mbs); /* from inverting diagonal blocks */ 1437 PetscFunctionReturn(0); 1438 } 1439 /* 1440 Version for when blocks are 5 by 5 Using natural ordering 1441 */ 1442 #undef __FUNC__ 1443 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_5_NaturalOrdering" 1444 int MatLUFactorNumeric_SeqSBAIJ_5_NaturalOrdering(Mat A,Mat *B) 1445 { 1446 Mat C = *B; 1447 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 1448 int ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 1449 int *ajtmpold,*ajtmp,nz,row; 1450 int *diag_offset = b->diag,*ai=a->i,*aj=a->j,*pj; 1451 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 1452 MatScalar x1,x2,x3,x4,x5,x6,x7,x8,x9,x10,x11,x12,x13,x14,x15; 1453 MatScalar x16,x17,x18,x19,x20,x21,x22,x23,x24,x25; 1454 MatScalar p1,p2,p3,p4,p5,p6,p7,p8,p9,p10,p11,p12,p13,p14,p15; 1455 MatScalar p16,p17,p18,p19,p20,p21,p22,p23,p24,p25; 1456 MatScalar m1,m2,m3,m4,m5,m6,m7,m8,m9,m10,m11,m12,m13,m14,m15; 1457 MatScalar m16,m17,m18,m19,m20,m21,m22,m23,m24,m25; 1458 MatScalar *ba = b->a,*aa = a->a; 1459 1460 PetscFunctionBegin; 1461 rtmp = (MatScalar*)PetscMalloc(25*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 1462 for (i=0; i<n; i++) { 1463 nz = bi[i+1] - bi[i]; 1464 ajtmp = bj + bi[i]; 1465 for (j=0; j<nz; j++) { 1466 x = rtmp+25*ajtmp[j]; 1467 x[0] = x[1] = x[2] = x[3] = x[4] = x[5] = x[6] = x[7] = x[8] = x[9] = 0.0; 1468 x[10] = x[11] = x[12] = x[13] = x[14] = x[15] = 0.0; 1469 x[16] = x[17] = x[18] = x[19] = x[20] = x[21] = x[22] = x[23] = x[24] = 0.0; 1470 } 1471 /* load in initial (unfactored row) */ 1472 nz = ai[i+1] - ai[i]; 1473 ajtmpold = aj + ai[i]; 1474 v = aa + 25*ai[i]; 1475 for (j=0; j<nz; j++) { 1476 x = rtmp+25*ajtmpold[j]; 1477 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 1478 x[4] = v[4]; x[5] = v[5]; x[6] = v[6]; x[7] = v[7]; x[8] = v[8]; 1479 x[9] = v[9]; x[10] = v[10]; x[11] = v[11]; x[12] = v[12]; x[13] = v[13]; 1480 x[14] = v[14]; x[15] = v[15]; x[16] = v[16]; x[17] = v[17]; x[18] = v[18]; 1481 x[19] = v[19]; x[20] = v[20]; x[21] = v[21]; x[22] = v[22]; x[23] = v[23]; 1482 x[24] = v[24]; 1483 v += 25; 1484 } 1485 row = *ajtmp++; 1486 while (row < i) { 1487 pc = rtmp + 25*row; 1488 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 1489 p5 = pc[4]; p6 = pc[5]; p7 = pc[6]; p8 = pc[7]; p9 = pc[8]; 1490 p10 = pc[9]; p11 = pc[10]; p12 = pc[11]; p13 = pc[12]; p14 = pc[13]; 1491 p15 = pc[14]; p16 = pc[15]; p17 = pc[16]; p18 = pc[17]; 1492 p19 = pc[18]; p20 = pc[19]; p21 = pc[20]; p22 = pc[21]; p23 = pc[22]; 1493 p24 = pc[23]; p25 = pc[24]; 1494 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0 || p5 != 0.0 || 1495 p6 != 0.0 || p7 != 0.0 || p8 != 0.0 || p9 != 0.0 || p10 != 0.0 || 1496 p11 != 0.0 || p12 != 0.0 || p13 != 0.0 || p14 != 0.0 || p15 != 0.0 1497 || p16 != 0.0 || p17 != 0.0 || p18 != 0.0 || p19 != 0.0 || p20 != 0.0 1498 || p21 != 0.0 || p22 != 0.0 || p23 != 0.0 || p24 != 0.0 || p25 != 0.0) { 1499 pv = ba + 25*diag_offset[row]; 1500 pj = bj + diag_offset[row] + 1; 1501 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1502 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 1503 x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; x13 = pv[12]; x14 = pv[13]; 1504 x15 = pv[14]; x16 = pv[15]; x17 = pv[16]; x18 = pv[17]; x19 = pv[18]; 1505 x20 = pv[19]; x21 = pv[20]; x22 = pv[21]; x23 = pv[22]; x24 = pv[23]; 1506 x25 = pv[24]; 1507 pc[0] = m1 = p1*x1 + p6*x2 + p11*x3 + p16*x4 + p21*x5; 1508 pc[1] = m2 = p2*x1 + p7*x2 + p12*x3 + p17*x4 + p22*x5; 1509 pc[2] = m3 = p3*x1 + p8*x2 + p13*x3 + p18*x4 + p23*x5; 1510 pc[3] = m4 = p4*x1 + p9*x2 + p14*x3 + p19*x4 + p24*x5; 1511 pc[4] = m5 = p5*x1 + p10*x2 + p15*x3 + p20*x4 + p25*x5; 1512 1513 pc[5] = m6 = p1*x6 + p6*x7 + p11*x8 + p16*x9 + p21*x10; 1514 pc[6] = m7 = p2*x6 + p7*x7 + p12*x8 + p17*x9 + p22*x10; 1515 pc[7] = m8 = p3*x6 + p8*x7 + p13*x8 + p18*x9 + p23*x10; 1516 pc[8] = m9 = p4*x6 + p9*x7 + p14*x8 + p19*x9 + p24*x10; 1517 pc[9] = m10 = p5*x6 + p10*x7 + p15*x8 + p20*x9 + p25*x10; 1518 1519 pc[10] = m11 = p1*x11 + p6*x12 + p11*x13 + p16*x14 + p21*x15; 1520 pc[11] = m12 = p2*x11 + p7*x12 + p12*x13 + p17*x14 + p22*x15; 1521 pc[12] = m13 = p3*x11 + p8*x12 + p13*x13 + p18*x14 + p23*x15; 1522 pc[13] = m14 = p4*x11 + p9*x12 + p14*x13 + p19*x14 + p24*x15; 1523 pc[14] = m15 = p5*x11 + p10*x12 + p15*x13 + p20*x14 + p25*x15; 1524 1525 pc[15] = m16 = p1*x16 + p6*x17 + p11*x18 + p16*x19 + p21*x20; 1526 pc[16] = m17 = p2*x16 + p7*x17 + p12*x18 + p17*x19 + p22*x20; 1527 pc[17] = m18 = p3*x16 + p8*x17 + p13*x18 + p18*x19 + p23*x20; 1528 pc[18] = m19 = p4*x16 + p9*x17 + p14*x18 + p19*x19 + p24*x20; 1529 pc[19] = m20 = p5*x16 + p10*x17 + p15*x18 + p20*x19 + p25*x20; 1530 1531 pc[20] = m21 = p1*x21 + p6*x22 + p11*x23 + p16*x24 + p21*x25; 1532 pc[21] = m22 = p2*x21 + p7*x22 + p12*x23 + p17*x24 + p22*x25; 1533 pc[22] = m23 = p3*x21 + p8*x22 + p13*x23 + p18*x24 + p23*x25; 1534 pc[23] = m24 = p4*x21 + p9*x22 + p14*x23 + p19*x24 + p24*x25; 1535 pc[24] = m25 = p5*x21 + p10*x22 + p15*x23 + p20*x24 + p25*x25; 1536 1537 nz = bi[row+1] - diag_offset[row] - 1; 1538 pv += 25; 1539 for (j=0; j<nz; j++) { 1540 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1541 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 1542 x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; x13 = pv[12]; 1543 x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; x17 = pv[16]; x18 = pv[17]; 1544 x19 = pv[18]; x20 = pv[19]; x21 = pv[20]; x22 = pv[21]; x23 = pv[22]; 1545 x24 = pv[23]; x25 = pv[24]; 1546 x = rtmp + 25*pj[j]; 1547 x[0] -= m1*x1 + m6*x2 + m11*x3 + m16*x4 + m21*x5; 1548 x[1] -= m2*x1 + m7*x2 + m12*x3 + m17*x4 + m22*x5; 1549 x[2] -= m3*x1 + m8*x2 + m13*x3 + m18*x4 + m23*x5; 1550 x[3] -= m4*x1 + m9*x2 + m14*x3 + m19*x4 + m24*x5; 1551 x[4] -= m5*x1 + m10*x2 + m15*x3 + m20*x4 + m25*x5; 1552 1553 x[5] -= m1*x6 + m6*x7 + m11*x8 + m16*x9 + m21*x10; 1554 x[6] -= m2*x6 + m7*x7 + m12*x8 + m17*x9 + m22*x10; 1555 x[7] -= m3*x6 + m8*x7 + m13*x8 + m18*x9 + m23*x10; 1556 x[8] -= m4*x6 + m9*x7 + m14*x8 + m19*x9 + m24*x10; 1557 x[9] -= m5*x6 + m10*x7 + m15*x8 + m20*x9 + m25*x10; 1558 1559 x[10] -= m1*x11 + m6*x12 + m11*x13 + m16*x14 + m21*x15; 1560 x[11] -= m2*x11 + m7*x12 + m12*x13 + m17*x14 + m22*x15; 1561 x[12] -= m3*x11 + m8*x12 + m13*x13 + m18*x14 + m23*x15; 1562 x[13] -= m4*x11 + m9*x12 + m14*x13 + m19*x14 + m24*x15; 1563 x[14] -= m5*x11 + m10*x12 + m15*x13 + m20*x14 + m25*x15; 1564 1565 x[15] -= m1*x16 + m6*x17 + m11*x18 + m16*x19 + m21*x20; 1566 x[16] -= m2*x16 + m7*x17 + m12*x18 + m17*x19 + m22*x20; 1567 x[17] -= m3*x16 + m8*x17 + m13*x18 + m18*x19 + m23*x20; 1568 x[18] -= m4*x16 + m9*x17 + m14*x18 + m19*x19 + m24*x20; 1569 x[19] -= m5*x16 + m10*x17 + m15*x18 + m20*x19 + m25*x20; 1570 1571 x[20] -= m1*x21 + m6*x22 + m11*x23 + m16*x24 + m21*x25; 1572 x[21] -= m2*x21 + m7*x22 + m12*x23 + m17*x24 + m22*x25; 1573 x[22] -= m3*x21 + m8*x22 + m13*x23 + m18*x24 + m23*x25; 1574 x[23] -= m4*x21 + m9*x22 + m14*x23 + m19*x24 + m24*x25; 1575 x[24] -= m5*x21 + m10*x22 + m15*x23 + m20*x24 + m25*x25; 1576 pv += 25; 1577 } 1578 PLogFlops(250*nz+225); 1579 } 1580 row = *ajtmp++; 1581 } 1582 /* finished row so stick it into b->a */ 1583 pv = ba + 25*bi[i]; 1584 pj = bj + bi[i]; 1585 nz = bi[i+1] - bi[i]; 1586 for (j=0; j<nz; j++) { 1587 x = rtmp+25*pj[j]; 1588 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 1589 pv[4] = x[4]; pv[5] = x[5]; pv[6] = x[6]; pv[7] = x[7]; pv[8] = x[8]; 1590 pv[9] = x[9]; pv[10] = x[10]; pv[11] = x[11]; pv[12] = x[12]; 1591 pv[13] = x[13]; pv[14] = x[14]; pv[15] = x[15]; pv[16] = x[16]; pv[17] = x[17]; 1592 pv[18] = x[18]; pv[19] = x[19]; pv[20] = x[20]; pv[21] = x[21]; pv[22] = x[22]; 1593 pv[23] = x[23]; pv[24] = x[24]; 1594 pv += 25; 1595 } 1596 /* invert diagonal block */ 1597 w = ba + 25*diag_offset[i]; 1598 ierr = Kernel_A_gets_inverse_A_5(w);CHKERRQ(ierr); 1599 } 1600 1601 ierr = PetscFree(rtmp);CHKERRQ(ierr); 1602 C->factor = FACTOR_LU; 1603 C->assembled = PETSC_TRUE; 1604 PLogFlops(1.3333*125*b->mbs); /* from inverting diagonal blocks */ 1605 PetscFunctionReturn(0); 1606 } 1607 1608 /* ------------------------------------------------------------*/ 1609 /* 1610 Version for when blocks are 4 by 4 1611 */ 1612 #undef __FUNC__ 1613 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_4" 1614 int MatLUFactorNumeric_SeqSBAIJ_4(Mat A,Mat *B) 1615 { 1616 Mat C = *B; 1617 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 1618 IS isrow = b->row,isicol = b->icol; 1619 int *r,*ic,ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 1620 int *ajtmpold,*ajtmp,nz,row; 1621 int *diag_offset = b->diag,idx,*ai=a->i,*aj=a->j,*pj; 1622 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 1623 MatScalar p1,p2,p3,p4,m1,m2,m3,m4,m5,m6,m7,m8,m9,x1,x2,x3,x4; 1624 MatScalar p5,p6,p7,p8,p9,x5,x6,x7,x8,x9,x10,x11,x12,x13,x14,x15,x16; 1625 MatScalar p10,p11,p12,p13,p14,p15,p16,m10,m11,m12; 1626 MatScalar m13,m14,m15,m16; 1627 MatScalar *ba = b->a,*aa = a->a; 1628 1629 PetscFunctionBegin; 1630 ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr); 1631 ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr); 1632 rtmp = (MatScalar*)PetscMalloc(16*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 1633 1634 for (i=0; i<n; i++) { 1635 nz = bi[i+1] - bi[i]; 1636 ajtmp = bj + bi[i]; 1637 for (j=0; j<nz; j++) { 1638 x = rtmp+16*ajtmp[j]; 1639 x[0] = x[1] = x[2] = x[3] = x[4] = x[5] = x[6] = x[7] = x[8] = x[9] = 0.0; 1640 x[10] = x[11] = x[12] = x[13] = x[14] = x[15] = 0.0; 1641 } 1642 /* load in initial (unfactored row) */ 1643 idx = r[i]; 1644 nz = ai[idx+1] - ai[idx]; 1645 ajtmpold = aj + ai[idx]; 1646 v = aa + 16*ai[idx]; 1647 for (j=0; j<nz; j++) { 1648 x = rtmp+16*ic[ajtmpold[j]]; 1649 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 1650 x[4] = v[4]; x[5] = v[5]; x[6] = v[6]; x[7] = v[7]; x[8] = v[8]; 1651 x[9] = v[9]; x[10] = v[10]; x[11] = v[11]; x[12] = v[12]; x[13] = v[13]; 1652 x[14] = v[14]; x[15] = v[15]; 1653 v += 16; 1654 } 1655 row = *ajtmp++; 1656 while (row < i) { 1657 pc = rtmp + 16*row; 1658 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 1659 p5 = pc[4]; p6 = pc[5]; p7 = pc[6]; p8 = pc[7]; p9 = pc[8]; 1660 p10 = pc[9]; p11 = pc[10]; p12 = pc[11]; p13 = pc[12]; p14 = pc[13]; 1661 p15 = pc[14]; p16 = pc[15]; 1662 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0 || p5 != 0.0 || 1663 p6 != 0.0 || p7 != 0.0 || p8 != 0.0 || p9 != 0.0 || p10 != 0.0 || 1664 p11 != 0.0 || p12 != 0.0 || p13 != 0.0 || p14 != 0.0 || p15 != 0.0 1665 || p16 != 0.0) { 1666 pv = ba + 16*diag_offset[row]; 1667 pj = bj + diag_offset[row] + 1; 1668 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1669 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 1670 x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; x13 = pv[12]; x14 = pv[13]; 1671 x15 = pv[14]; x16 = pv[15]; 1672 pc[0] = m1 = p1*x1 + p5*x2 + p9*x3 + p13*x4; 1673 pc[1] = m2 = p2*x1 + p6*x2 + p10*x3 + p14*x4; 1674 pc[2] = m3 = p3*x1 + p7*x2 + p11*x3 + p15*x4; 1675 pc[3] = m4 = p4*x1 + p8*x2 + p12*x3 + p16*x4; 1676 1677 pc[4] = m5 = p1*x5 + p5*x6 + p9*x7 + p13*x8; 1678 pc[5] = m6 = p2*x5 + p6*x6 + p10*x7 + p14*x8; 1679 pc[6] = m7 = p3*x5 + p7*x6 + p11*x7 + p15*x8; 1680 pc[7] = m8 = p4*x5 + p8*x6 + p12*x7 + p16*x8; 1681 1682 pc[8] = m9 = p1*x9 + p5*x10 + p9*x11 + p13*x12; 1683 pc[9] = m10 = p2*x9 + p6*x10 + p10*x11 + p14*x12; 1684 pc[10] = m11 = p3*x9 + p7*x10 + p11*x11 + p15*x12; 1685 pc[11] = m12 = p4*x9 + p8*x10 + p12*x11 + p16*x12; 1686 1687 pc[12] = m13 = p1*x13 + p5*x14 + p9*x15 + p13*x16; 1688 pc[13] = m14 = p2*x13 + p6*x14 + p10*x15 + p14*x16; 1689 pc[14] = m15 = p3*x13 + p7*x14 + p11*x15 + p15*x16; 1690 pc[15] = m16 = p4*x13 + p8*x14 + p12*x15 + p16*x16; 1691 1692 nz = bi[row+1] - diag_offset[row] - 1; 1693 pv += 16; 1694 for (j=0; j<nz; j++) { 1695 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1696 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 1697 x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; x13 = pv[12]; 1698 x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; 1699 x = rtmp + 16*pj[j]; 1700 x[0] -= m1*x1 + m5*x2 + m9*x3 + m13*x4; 1701 x[1] -= m2*x1 + m6*x2 + m10*x3 + m14*x4; 1702 x[2] -= m3*x1 + m7*x2 + m11*x3 + m15*x4; 1703 x[3] -= m4*x1 + m8*x2 + m12*x3 + m16*x4; 1704 1705 x[4] -= m1*x5 + m5*x6 + m9*x7 + m13*x8; 1706 x[5] -= m2*x5 + m6*x6 + m10*x7 + m14*x8; 1707 x[6] -= m3*x5 + m7*x6 + m11*x7 + m15*x8; 1708 x[7] -= m4*x5 + m8*x6 + m12*x7 + m16*x8; 1709 1710 x[8] -= m1*x9 + m5*x10 + m9*x11 + m13*x12; 1711 x[9] -= m2*x9 + m6*x10 + m10*x11 + m14*x12; 1712 x[10] -= m3*x9 + m7*x10 + m11*x11 + m15*x12; 1713 x[11] -= m4*x9 + m8*x10 + m12*x11 + m16*x12; 1714 1715 x[12] -= m1*x13 + m5*x14 + m9*x15 + m13*x16; 1716 x[13] -= m2*x13 + m6*x14 + m10*x15 + m14*x16; 1717 x[14] -= m3*x13 + m7*x14 + m11*x15 + m15*x16; 1718 x[15] -= m4*x13 + m8*x14 + m12*x15 + m16*x16; 1719 1720 pv += 16; 1721 } 1722 PLogFlops(128*nz+112); 1723 } 1724 row = *ajtmp++; 1725 } 1726 /* finished row so stick it into b->a */ 1727 pv = ba + 16*bi[i]; 1728 pj = bj + bi[i]; 1729 nz = bi[i+1] - bi[i]; 1730 for (j=0; j<nz; j++) { 1731 x = rtmp+16*pj[j]; 1732 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 1733 pv[4] = x[4]; pv[5] = x[5]; pv[6] = x[6]; pv[7] = x[7]; pv[8] = x[8]; 1734 pv[9] = x[9]; pv[10] = x[10]; pv[11] = x[11]; pv[12] = x[12]; 1735 pv[13] = x[13]; pv[14] = x[14]; pv[15] = x[15]; 1736 pv += 16; 1737 } 1738 /* invert diagonal block */ 1739 w = ba + 16*diag_offset[i]; 1740 ierr = Kernel_A_gets_inverse_A_4(w);CHKERRQ(ierr); 1741 } 1742 1743 ierr = PetscFree(rtmp);CHKERRQ(ierr); 1744 ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr); 1745 ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr); 1746 C->factor = FACTOR_LU; 1747 C->assembled = PETSC_TRUE; 1748 PLogFlops(1.3333*64*b->mbs); /* from inverting diagonal blocks */ 1749 PetscFunctionReturn(0); 1750 } 1751 /* 1752 Version for when blocks are 4 by 4 Using natural ordering 1753 */ 1754 #undef __FUNC__ 1755 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_4_NaturalOrdering" 1756 int MatLUFactorNumeric_SeqSBAIJ_4_NaturalOrdering(Mat A,Mat *B) 1757 { 1758 Mat C = *B; 1759 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 1760 int ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 1761 int *ajtmpold,*ajtmp,nz,row; 1762 int *diag_offset = b->diag,*ai=a->i,*aj=a->j,*pj; 1763 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 1764 MatScalar p1,p2,p3,p4,m1,m2,m3,m4,m5,m6,m7,m8,m9,x1,x2,x3,x4; 1765 MatScalar p5,p6,p7,p8,p9,x5,x6,x7,x8,x9,x10,x11,x12,x13,x14,x15,x16; 1766 MatScalar p10,p11,p12,p13,p14,p15,p16,m10,m11,m12; 1767 MatScalar m13,m14,m15,m16; 1768 MatScalar *ba = b->a,*aa = a->a; 1769 1770 PetscFunctionBegin; 1771 rtmp = (MatScalar*)PetscMalloc(16*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 1772 1773 for (i=0; i<n; i++) { 1774 nz = bi[i+1] - bi[i]; 1775 ajtmp = bj + bi[i]; 1776 for (j=0; j<nz; j++) { 1777 x = rtmp+16*ajtmp[j]; 1778 x[0] = x[1] = x[2] = x[3] = x[4] = x[5] = x[6] = x[7] = x[8] = x[9] = 0.0; 1779 x[10] = x[11] = x[12] = x[13] = x[14] = x[15] = 0.0; 1780 } 1781 /* load in initial (unfactored row) */ 1782 nz = ai[i+1] - ai[i]; 1783 ajtmpold = aj + ai[i]; 1784 v = aa + 16*ai[i]; 1785 for (j=0; j<nz; j++) { 1786 x = rtmp+16*ajtmpold[j]; 1787 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 1788 x[4] = v[4]; x[5] = v[5]; x[6] = v[6]; x[7] = v[7]; x[8] = v[8]; 1789 x[9] = v[9]; x[10] = v[10]; x[11] = v[11]; x[12] = v[12]; x[13] = v[13]; 1790 x[14] = v[14]; x[15] = v[15]; 1791 v += 16; 1792 } 1793 row = *ajtmp++; 1794 while (row < i) { 1795 pc = rtmp + 16*row; 1796 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 1797 p5 = pc[4]; p6 = pc[5]; p7 = pc[6]; p8 = pc[7]; p9 = pc[8]; 1798 p10 = pc[9]; p11 = pc[10]; p12 = pc[11]; p13 = pc[12]; p14 = pc[13]; 1799 p15 = pc[14]; p16 = pc[15]; 1800 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0 || p5 != 0.0 || 1801 p6 != 0.0 || p7 != 0.0 || p8 != 0.0 || p9 != 0.0 || p10 != 0.0 || 1802 p11 != 0.0 || p12 != 0.0 || p13 != 0.0 || p14 != 0.0 || p15 != 0.0 1803 || p16 != 0.0) { 1804 pv = ba + 16*diag_offset[row]; 1805 pj = bj + diag_offset[row] + 1; 1806 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1807 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 1808 x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; x13 = pv[12]; x14 = pv[13]; 1809 x15 = pv[14]; x16 = pv[15]; 1810 pc[0] = m1 = p1*x1 + p5*x2 + p9*x3 + p13*x4; 1811 pc[1] = m2 = p2*x1 + p6*x2 + p10*x3 + p14*x4; 1812 pc[2] = m3 = p3*x1 + p7*x2 + p11*x3 + p15*x4; 1813 pc[3] = m4 = p4*x1 + p8*x2 + p12*x3 + p16*x4; 1814 1815 pc[4] = m5 = p1*x5 + p5*x6 + p9*x7 + p13*x8; 1816 pc[5] = m6 = p2*x5 + p6*x6 + p10*x7 + p14*x8; 1817 pc[6] = m7 = p3*x5 + p7*x6 + p11*x7 + p15*x8; 1818 pc[7] = m8 = p4*x5 + p8*x6 + p12*x7 + p16*x8; 1819 1820 pc[8] = m9 = p1*x9 + p5*x10 + p9*x11 + p13*x12; 1821 pc[9] = m10 = p2*x9 + p6*x10 + p10*x11 + p14*x12; 1822 pc[10] = m11 = p3*x9 + p7*x10 + p11*x11 + p15*x12; 1823 pc[11] = m12 = p4*x9 + p8*x10 + p12*x11 + p16*x12; 1824 1825 pc[12] = m13 = p1*x13 + p5*x14 + p9*x15 + p13*x16; 1826 pc[13] = m14 = p2*x13 + p6*x14 + p10*x15 + p14*x16; 1827 pc[14] = m15 = p3*x13 + p7*x14 + p11*x15 + p15*x16; 1828 pc[15] = m16 = p4*x13 + p8*x14 + p12*x15 + p16*x16; 1829 1830 nz = bi[row+1] - diag_offset[row] - 1; 1831 pv += 16; 1832 for (j=0; j<nz; j++) { 1833 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1834 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 1835 x10 = pv[9]; x11 = pv[10]; x12 = pv[11]; x13 = pv[12]; 1836 x14 = pv[13]; x15 = pv[14]; x16 = pv[15]; 1837 x = rtmp + 16*pj[j]; 1838 x[0] -= m1*x1 + m5*x2 + m9*x3 + m13*x4; 1839 x[1] -= m2*x1 + m6*x2 + m10*x3 + m14*x4; 1840 x[2] -= m3*x1 + m7*x2 + m11*x3 + m15*x4; 1841 x[3] -= m4*x1 + m8*x2 + m12*x3 + m16*x4; 1842 1843 x[4] -= m1*x5 + m5*x6 + m9*x7 + m13*x8; 1844 x[5] -= m2*x5 + m6*x6 + m10*x7 + m14*x8; 1845 x[6] -= m3*x5 + m7*x6 + m11*x7 + m15*x8; 1846 x[7] -= m4*x5 + m8*x6 + m12*x7 + m16*x8; 1847 1848 x[8] -= m1*x9 + m5*x10 + m9*x11 + m13*x12; 1849 x[9] -= m2*x9 + m6*x10 + m10*x11 + m14*x12; 1850 x[10] -= m3*x9 + m7*x10 + m11*x11 + m15*x12; 1851 x[11] -= m4*x9 + m8*x10 + m12*x11 + m16*x12; 1852 1853 x[12] -= m1*x13 + m5*x14 + m9*x15 + m13*x16; 1854 x[13] -= m2*x13 + m6*x14 + m10*x15 + m14*x16; 1855 x[14] -= m3*x13 + m7*x14 + m11*x15 + m15*x16; 1856 x[15] -= m4*x13 + m8*x14 + m12*x15 + m16*x16; 1857 1858 pv += 16; 1859 } 1860 PLogFlops(128*nz+112); 1861 } 1862 row = *ajtmp++; 1863 } 1864 /* finished row so stick it into b->a */ 1865 pv = ba + 16*bi[i]; 1866 pj = bj + bi[i]; 1867 nz = bi[i+1] - bi[i]; 1868 for (j=0; j<nz; j++) { 1869 x = rtmp+16*pj[j]; 1870 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 1871 pv[4] = x[4]; pv[5] = x[5]; pv[6] = x[6]; pv[7] = x[7]; pv[8] = x[8]; 1872 pv[9] = x[9]; pv[10] = x[10]; pv[11] = x[11]; pv[12] = x[12]; 1873 pv[13] = x[13]; pv[14] = x[14]; pv[15] = x[15]; 1874 pv += 16; 1875 } 1876 /* invert diagonal block */ 1877 w = ba + 16*diag_offset[i]; 1878 ierr = Kernel_A_gets_inverse_A_4(w);CHKERRQ(ierr); 1879 } 1880 1881 ierr = PetscFree(rtmp);CHKERRQ(ierr); 1882 C->factor = FACTOR_LU; 1883 C->assembled = PETSC_TRUE; 1884 PLogFlops(1.3333*64*b->mbs); /* from inverting diagonal blocks */ 1885 PetscFunctionReturn(0); 1886 } 1887 1888 1889 /* ------------------------------------------------------------*/ 1890 /* 1891 Version for when blocks are 3 by 3 1892 */ 1893 #undef __FUNC__ 1894 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_3" 1895 int MatLUFactorNumeric_SeqSBAIJ_3(Mat A,Mat *B) 1896 { 1897 Mat C = *B; 1898 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 1899 IS isrow = b->row,isicol = b->icol; 1900 int *r,*ic,ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 1901 int *ajtmpold,*ajtmp,nz,row,*ai=a->i,*aj=a->j; 1902 int *diag_offset = b->diag,idx,*pj; 1903 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 1904 MatScalar p1,p2,p3,p4,m1,m2,m3,m4,m5,m6,m7,m8,m9,x1,x2,x3,x4; 1905 MatScalar p5,p6,p7,p8,p9,x5,x6,x7,x8,x9; 1906 MatScalar *ba = b->a,*aa = a->a; 1907 1908 PetscFunctionBegin; 1909 ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr); 1910 ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr); 1911 rtmp = (MatScalar*)PetscMalloc(9*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 1912 1913 for (i=0; i<n; i++) { 1914 nz = bi[i+1] - bi[i]; 1915 ajtmp = bj + bi[i]; 1916 for (j=0; j<nz; j++) { 1917 x = rtmp + 9*ajtmp[j]; 1918 x[0] = x[1] = x[2] = x[3] = x[4] = x[5] = x[6] = x[7] = x[8] = 0.0; 1919 } 1920 /* load in initial (unfactored row) */ 1921 idx = r[i]; 1922 nz = ai[idx+1] - ai[idx]; 1923 ajtmpold = aj + ai[idx]; 1924 v = aa + 9*ai[idx]; 1925 for (j=0; j<nz; j++) { 1926 x = rtmp + 9*ic[ajtmpold[j]]; 1927 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 1928 x[4] = v[4]; x[5] = v[5]; x[6] = v[6]; x[7] = v[7]; x[8] = v[8]; 1929 v += 9; 1930 } 1931 row = *ajtmp++; 1932 while (row < i) { 1933 pc = rtmp + 9*row; 1934 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 1935 p5 = pc[4]; p6 = pc[5]; p7 = pc[6]; p8 = pc[7]; p9 = pc[8]; 1936 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0 || p5 != 0.0 || 1937 p6 != 0.0 || p7 != 0.0 || p8 != 0.0 || p9 != 0.0) { 1938 pv = ba + 9*diag_offset[row]; 1939 pj = bj + diag_offset[row] + 1; 1940 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1941 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 1942 pc[0] = m1 = p1*x1 + p4*x2 + p7*x3; 1943 pc[1] = m2 = p2*x1 + p5*x2 + p8*x3; 1944 pc[2] = m3 = p3*x1 + p6*x2 + p9*x3; 1945 1946 pc[3] = m4 = p1*x4 + p4*x5 + p7*x6; 1947 pc[4] = m5 = p2*x4 + p5*x5 + p8*x6; 1948 pc[5] = m6 = p3*x4 + p6*x5 + p9*x6; 1949 1950 pc[6] = m7 = p1*x7 + p4*x8 + p7*x9; 1951 pc[7] = m8 = p2*x7 + p5*x8 + p8*x9; 1952 pc[8] = m9 = p3*x7 + p6*x8 + p9*x9; 1953 nz = bi[row+1] - diag_offset[row] - 1; 1954 pv += 9; 1955 for (j=0; j<nz; j++) { 1956 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 1957 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 1958 x = rtmp + 9*pj[j]; 1959 x[0] -= m1*x1 + m4*x2 + m7*x3; 1960 x[1] -= m2*x1 + m5*x2 + m8*x3; 1961 x[2] -= m3*x1 + m6*x2 + m9*x3; 1962 1963 x[3] -= m1*x4 + m4*x5 + m7*x6; 1964 x[4] -= m2*x4 + m5*x5 + m8*x6; 1965 x[5] -= m3*x4 + m6*x5 + m9*x6; 1966 1967 x[6] -= m1*x7 + m4*x8 + m7*x9; 1968 x[7] -= m2*x7 + m5*x8 + m8*x9; 1969 x[8] -= m3*x7 + m6*x8 + m9*x9; 1970 pv += 9; 1971 } 1972 PLogFlops(54*nz+36); 1973 } 1974 row = *ajtmp++; 1975 } 1976 /* finished row so stick it into b->a */ 1977 pv = ba + 9*bi[i]; 1978 pj = bj + bi[i]; 1979 nz = bi[i+1] - bi[i]; 1980 for (j=0; j<nz; j++) { 1981 x = rtmp + 9*pj[j]; 1982 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 1983 pv[4] = x[4]; pv[5] = x[5]; pv[6] = x[6]; pv[7] = x[7]; pv[8] = x[8]; 1984 pv += 9; 1985 } 1986 /* invert diagonal block */ 1987 w = ba + 9*diag_offset[i]; 1988 ierr = Kernel_A_gets_inverse_A_3(w);CHKERRQ(ierr); 1989 } 1990 1991 ierr = PetscFree(rtmp);CHKERRQ(ierr); 1992 ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr); 1993 ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr); 1994 C->factor = FACTOR_LU; 1995 C->assembled = PETSC_TRUE; 1996 PLogFlops(1.3333*27*b->mbs); /* from inverting diagonal blocks */ 1997 PetscFunctionReturn(0); 1998 } 1999 /* 2000 Version for when blocks are 3 by 3 Using natural ordering 2001 */ 2002 #undef __FUNC__ 2003 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_3_NaturalOrdering" 2004 int MatLUFactorNumeric_SeqSBAIJ_3_NaturalOrdering(Mat A,Mat *B) 2005 { 2006 Mat C = *B; 2007 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 2008 int ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 2009 int *ajtmpold,*ajtmp,nz,row; 2010 int *diag_offset = b->diag,*ai=a->i,*aj=a->j,*pj; 2011 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 2012 MatScalar p1,p2,p3,p4,m1,m2,m3,m4,m5,m6,m7,m8,m9,x1,x2,x3,x4; 2013 MatScalar p5,p6,p7,p8,p9,x5,x6,x7,x8,x9; 2014 MatScalar *ba = b->a,*aa = a->a; 2015 2016 PetscFunctionBegin; 2017 rtmp = (MatScalar*)PetscMalloc(9*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 2018 2019 for (i=0; i<n; i++) { 2020 nz = bi[i+1] - bi[i]; 2021 ajtmp = bj + bi[i]; 2022 for (j=0; j<nz; j++) { 2023 x = rtmp+9*ajtmp[j]; 2024 x[0] = x[1] = x[2] = x[3] = x[4] = x[5] = x[6] = x[7] = x[8] = 0.0; 2025 } 2026 /* load in initial (unfactored row) */ 2027 nz = ai[i+1] - ai[i]; 2028 ajtmpold = aj + ai[i]; 2029 v = aa + 9*ai[i]; 2030 for (j=0; j<nz; j++) { 2031 x = rtmp+9*ajtmpold[j]; 2032 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 2033 x[4] = v[4]; x[5] = v[5]; x[6] = v[6]; x[7] = v[7]; x[8] = v[8]; 2034 v += 9; 2035 } 2036 row = *ajtmp++; 2037 while (row < i) { 2038 pc = rtmp + 9*row; 2039 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 2040 p5 = pc[4]; p6 = pc[5]; p7 = pc[6]; p8 = pc[7]; p9 = pc[8]; 2041 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0 || p5 != 0.0 || 2042 p6 != 0.0 || p7 != 0.0 || p8 != 0.0 || p9 != 0.0) { 2043 pv = ba + 9*diag_offset[row]; 2044 pj = bj + diag_offset[row] + 1; 2045 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 2046 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 2047 pc[0] = m1 = p1*x1 + p4*x2 + p7*x3; 2048 pc[1] = m2 = p2*x1 + p5*x2 + p8*x3; 2049 pc[2] = m3 = p3*x1 + p6*x2 + p9*x3; 2050 2051 pc[3] = m4 = p1*x4 + p4*x5 + p7*x6; 2052 pc[4] = m5 = p2*x4 + p5*x5 + p8*x6; 2053 pc[5] = m6 = p3*x4 + p6*x5 + p9*x6; 2054 2055 pc[6] = m7 = p1*x7 + p4*x8 + p7*x9; 2056 pc[7] = m8 = p2*x7 + p5*x8 + p8*x9; 2057 pc[8] = m9 = p3*x7 + p6*x8 + p9*x9; 2058 2059 nz = bi[row+1] - diag_offset[row] - 1; 2060 pv += 9; 2061 for (j=0; j<nz; j++) { 2062 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 2063 x5 = pv[4]; x6 = pv[5]; x7 = pv[6]; x8 = pv[7]; x9 = pv[8]; 2064 x = rtmp + 9*pj[j]; 2065 x[0] -= m1*x1 + m4*x2 + m7*x3; 2066 x[1] -= m2*x1 + m5*x2 + m8*x3; 2067 x[2] -= m3*x1 + m6*x2 + m9*x3; 2068 2069 x[3] -= m1*x4 + m4*x5 + m7*x6; 2070 x[4] -= m2*x4 + m5*x5 + m8*x6; 2071 x[5] -= m3*x4 + m6*x5 + m9*x6; 2072 2073 x[6] -= m1*x7 + m4*x8 + m7*x9; 2074 x[7] -= m2*x7 + m5*x8 + m8*x9; 2075 x[8] -= m3*x7 + m6*x8 + m9*x9; 2076 pv += 9; 2077 } 2078 PLogFlops(54*nz+36); 2079 } 2080 row = *ajtmp++; 2081 } 2082 /* finished row so stick it into b->a */ 2083 pv = ba + 9*bi[i]; 2084 pj = bj + bi[i]; 2085 nz = bi[i+1] - bi[i]; 2086 for (j=0; j<nz; j++) { 2087 x = rtmp+9*pj[j]; 2088 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 2089 pv[4] = x[4]; pv[5] = x[5]; pv[6] = x[6]; pv[7] = x[7]; pv[8] = x[8]; 2090 pv += 9; 2091 } 2092 /* invert diagonal block */ 2093 w = ba + 9*diag_offset[i]; 2094 ierr = Kernel_A_gets_inverse_A_3(w);CHKERRQ(ierr); 2095 } 2096 2097 ierr = PetscFree(rtmp);CHKERRQ(ierr); 2098 C->factor = FACTOR_LU; 2099 C->assembled = PETSC_TRUE; 2100 PLogFlops(1.3333*27*b->mbs); /* from inverting diagonal blocks */ 2101 PetscFunctionReturn(0); 2102 } 2103 2104 /* ------------------------------------------------------------*/ 2105 /* 2106 Version for when blocks are 2 by 2 2107 */ 2108 #undef __FUNC__ 2109 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_2" 2110 int MatLUFactorNumeric_SeqSBAIJ_2(Mat A,Mat *B) 2111 { 2112 Mat C = *B; 2113 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 2114 IS isrow = b->row,isicol = b->icol; 2115 int *r,*ic,ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 2116 int *ajtmpold,*ajtmp,nz,row; 2117 int *diag_offset=b->diag,idx,*ai=a->i,*aj=a->j,*pj; 2118 MatScalar *pv,*v,*rtmp,m1,m2,m3,m4,*pc,*w,*x,x1,x2,x3,x4; 2119 MatScalar p1,p2,p3,p4; 2120 MatScalar *ba = b->a,*aa = a->a; 2121 2122 PetscFunctionBegin; 2123 ierr = ISGetIndices(isrow,&r);CHKERRQ(ierr); 2124 ierr = ISGetIndices(isicol,&ic);CHKERRQ(ierr); 2125 rtmp = (MatScalar*)PetscMalloc(4*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 2126 2127 for (i=0; i<n; i++) { 2128 nz = bi[i+1] - bi[i]; 2129 ajtmp = bj + bi[i]; 2130 for (j=0; j<nz; j++) { 2131 x = rtmp+4*ajtmp[j]; x[0] = x[1] = x[2] = x[3] = 0.0; 2132 } 2133 /* load in initial (unfactored row) */ 2134 idx = r[i]; 2135 nz = ai[idx+1] - ai[idx]; 2136 ajtmpold = aj + ai[idx]; 2137 v = aa + 4*ai[idx]; 2138 for (j=0; j<nz; j++) { 2139 x = rtmp+4*ic[ajtmpold[j]]; 2140 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 2141 v += 4; 2142 } 2143 row = *ajtmp++; 2144 while (row < i) { 2145 pc = rtmp + 4*row; 2146 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 2147 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0) { 2148 pv = ba + 4*diag_offset[row]; 2149 pj = bj + diag_offset[row] + 1; 2150 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 2151 pc[0] = m1 = p1*x1 + p3*x2; 2152 pc[1] = m2 = p2*x1 + p4*x2; 2153 pc[2] = m3 = p1*x3 + p3*x4; 2154 pc[3] = m4 = p2*x3 + p4*x4; 2155 nz = bi[row+1] - diag_offset[row] - 1; 2156 pv += 4; 2157 for (j=0; j<nz; j++) { 2158 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 2159 x = rtmp + 4*pj[j]; 2160 x[0] -= m1*x1 + m3*x2; 2161 x[1] -= m2*x1 + m4*x2; 2162 x[2] -= m1*x3 + m3*x4; 2163 x[3] -= m2*x3 + m4*x4; 2164 pv += 4; 2165 } 2166 PLogFlops(16*nz+12); 2167 } 2168 row = *ajtmp++; 2169 } 2170 /* finished row so stick it into b->a */ 2171 pv = ba + 4*bi[i]; 2172 pj = bj + bi[i]; 2173 nz = bi[i+1] - bi[i]; 2174 for (j=0; j<nz; j++) { 2175 x = rtmp+4*pj[j]; 2176 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 2177 pv += 4; 2178 } 2179 /* invert diagonal block */ 2180 w = ba + 4*diag_offset[i]; 2181 ierr = Kernel_A_gets_inverse_A_2(w);CHKERRQ(ierr); 2182 /*Kernel_A_gets_inverse_A(bs,w,v_pivots,v_work);*/ 2183 } 2184 2185 ierr = PetscFree(rtmp);CHKERRQ(ierr); 2186 ierr = ISRestoreIndices(isicol,&ic);CHKERRQ(ierr); 2187 ierr = ISRestoreIndices(isrow,&r);CHKERRQ(ierr); 2188 C->factor = FACTOR_LU; 2189 C->assembled = PETSC_TRUE; 2190 PLogFlops(1.3333*8*b->mbs); /* from inverting diagonal blocks */ 2191 PetscFunctionReturn(0); 2192 } 2193 /* 2194 Version for when blocks are 2 by 2 Using natural ordering 2195 */ 2196 #undef __FUNC__ 2197 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_2_NaturalOrdering" 2198 int MatLUFactorNumeric_SeqSBAIJ_2_NaturalOrdering(Mat A,Mat *B) 2199 { 2200 Mat C = *B; 2201 Mat_SeqBAIJ *a = (Mat_SeqBAIJ*)A->data,*b = (Mat_SeqBAIJ *)C->data; 2202 int ierr,i,j,n = a->mbs,*bi = b->i,*bj = b->j; 2203 int *ajtmpold,*ajtmp,nz,row; 2204 int *diag_offset = b->diag,*ai=a->i,*aj=a->j,*pj; 2205 MatScalar *pv,*v,*rtmp,*pc,*w,*x; 2206 MatScalar p1,p2,p3,p4,m1,m2,m3,m4,x1,x2,x3,x4; 2207 MatScalar *ba = b->a,*aa = a->a; 2208 2209 PetscFunctionBegin; 2210 rtmp = (MatScalar*)PetscMalloc(4*(n+1)*sizeof(MatScalar));CHKPTRQ(rtmp); 2211 2212 for (i=0; i<n; i++) { 2213 nz = bi[i+1] - bi[i]; 2214 ajtmp = bj + bi[i]; 2215 for (j=0; j<nz; j++) { 2216 x = rtmp+4*ajtmp[j]; 2217 x[0] = x[1] = x[2] = x[3] = 0.0; 2218 } 2219 /* load in initial (unfactored row) */ 2220 nz = ai[i+1] - ai[i]; 2221 ajtmpold = aj + ai[i]; 2222 v = aa + 4*ai[i]; 2223 for (j=0; j<nz; j++) { 2224 x = rtmp+4*ajtmpold[j]; 2225 x[0] = v[0]; x[1] = v[1]; x[2] = v[2]; x[3] = v[3]; 2226 v += 4; 2227 } 2228 row = *ajtmp++; 2229 while (row < i) { 2230 pc = rtmp + 4*row; 2231 p1 = pc[0]; p2 = pc[1]; p3 = pc[2]; p4 = pc[3]; 2232 if (p1 != 0.0 || p2 != 0.0 || p3 != 0.0 || p4 != 0.0) { 2233 pv = ba + 4*diag_offset[row]; 2234 pj = bj + diag_offset[row] + 1; 2235 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 2236 pc[0] = m1 = p1*x1 + p3*x2; 2237 pc[1] = m2 = p2*x1 + p4*x2; 2238 pc[2] = m3 = p1*x3 + p3*x4; 2239 pc[3] = m4 = p2*x3 + p4*x4; 2240 nz = bi[row+1] - diag_offset[row] - 1; 2241 pv += 4; 2242 for (j=0; j<nz; j++) { 2243 x1 = pv[0]; x2 = pv[1]; x3 = pv[2]; x4 = pv[3]; 2244 x = rtmp + 4*pj[j]; 2245 x[0] -= m1*x1 + m3*x2; 2246 x[1] -= m2*x1 + m4*x2; 2247 x[2] -= m1*x3 + m3*x4; 2248 x[3] -= m2*x3 + m4*x4; 2249 pv += 4; 2250 } 2251 PLogFlops(16*nz+12); 2252 } 2253 row = *ajtmp++; 2254 } 2255 /* finished row so stick it into b->a */ 2256 pv = ba + 4*bi[i]; 2257 pj = bj + bi[i]; 2258 nz = bi[i+1] - bi[i]; 2259 for (j=0; j<nz; j++) { 2260 x = rtmp+4*pj[j]; 2261 pv[0] = x[0]; pv[1] = x[1]; pv[2] = x[2]; pv[3] = x[3]; 2262 pv += 4; 2263 } 2264 /* invert diagonal block */ 2265 w = ba + 4*diag_offset[i]; 2266 ierr = Kernel_A_gets_inverse_A_2(w);CHKERRQ(ierr); 2267 /*Kernel_A_gets_inverse_A(bs,w,v_pivots,v_work);*/ 2268 } 2269 2270 ierr = PetscFree(rtmp);CHKERRQ(ierr); 2271 C->factor = FACTOR_LU; 2272 C->assembled = PETSC_TRUE; 2273 PLogFlops(1.3333*8*b->mbs); /* from inverting diagonal blocks */ 2274 PetscFunctionReturn(0); 2275 } 2276 2277 /* ----------------------------------------------------------- */ 2278 /* 2279 Version for when blocks are 1 by 1. 2280 */ 2281 #undef __FUNC__ 2282 #define __FUNC__ "MatLUFactorNumeric_SeqSBAIJ_1" 2283 int MatLUFactorNumeric_SeqSBAIJ_1(Mat A,Mat *B) 2284 { 2285 Mat C = *B; 2286 Mat_SeqSBAIJ *a = (Mat_SeqSBAIJ*)A->data,*b = (Mat_SeqSBAIJ *)C->data; 2287 IS ip = b->row; 2288 int *rip,*riip,ierr,i,j,mbs = a->mbs,*bi = b->i,*bj = b->j; 2289 int *ai = a->i,*aj = a->j; 2290 MatScalar *rtmp; 2291 MatScalar *ba = b->a,*aa = a->a; 2292 MatScalar dk,uikdi; 2293 int k,jmin,jmax,*jl,*il,vj,nexti,juj,ili; 2294 2295 PetscFunctionBegin; 2296 ierr = ISGetIndices(ip,&rip);CHKERRQ(ierr); 2297 riip = rip; 2298 2299 /* INITIALIZATION */ 2300 /* il and jl record the first nonzero element in each row of the accessing 2301 window U(0:k, k:mbs-1). 2302 jl: list of rows to be added to uneliminated rows 2303 i>= k: jl(i) is the first row to be added to row i 2304 i< k: jl(i) is the row following row i in some list of rows 2305 jl(i) = mbs indicates the end of a list 2306 il(i): points to the first nonzero element in columns k,...,mbs-1 of 2307 row i of U */ 2308 rtmp = (MatScalar*)PetscMalloc(mbs*sizeof(MatScalar));CHKPTRQ(rtmp); 2309 il = (int*)PetscMalloc(mbs*sizeof(int));CHKPTRQ(il); 2310 jl = (int*)PetscMalloc(mbs*sizeof(int));CHKPTRQ(jl); 2311 for (i=0; i<mbs; i++) { 2312 rtmp[i] = 0.0; jl[i] = mbs; il[0] = 0; 2313 } 2314 2315 /* FOR EACH ROW K */ 2316 for (k = 0; k<mbs; k++){ 2317 2318 /* INITIALIZE K-TH ROW WITH ELEMENTS NONZERO IN ROW P(K) OF A */ 2319 jmin = ai[rip[k]]; jmax = ai[rip[k]+1]; 2320 if (jmin < jmax) { 2321 for (j = jmin; j < jmax; j++){ 2322 vj = riip[aj[j]]; 2323 if (k <= vj) rtmp[vj] = aa[j]; 2324 } 2325 } 2326 2327 /* MODIFY K-TH ROW BY ADDING IN THOSE ROWS I WITH U(I,K) NE 0 2328 FOR EACH ROW I TO BE ADDED IN */ 2329 dk = rtmp[k]; 2330 i = jl[k]; /* first row to be added to k_th row */ 2331 /* printf(" k=%d, pivot row = %d\n",k,i); */ 2332 2333 while (i < mbs){ 2334 nexti = jl[i]; /* next row to be added to k_th row */ 2335 /* printf(" pivot row = %d\n", nexti); */ 2336 2337 /* COMPUTE MULTIPLIER AND UPDATE DIAGONAL ELEMENT */ 2338 ili = il[i]; /* index of first nonzero element in U(i,k:bms-1) */ 2339 uikdi = - ba[ili]*ba[i]; 2340 dk += uikdi*ba[ili]; 2341 ba[ili] = uikdi; /* update U(i,k) */ 2342 2343 /* ADD MULTIPLE OF ROW I TO K-TH ROW ... */ 2344 jmin = ili + 1; jmax = bi[i+1]; 2345 if (jmin < jmax){ 2346 for (j=jmin; j<jmax; j++) rtmp[bj[j]] += uikdi*ba[j]; 2347 /* ... AND ADD I TO ROW LIST FOR NEXT NONZERO ENTRY */ 2348 il[i] = jmin; /* update il(i) in column k+1, ... mbs-1 */ 2349 j = bj[jmin]; 2350 jl[i] = jl[j]; jl[j] = i; /* update jl */ 2351 } 2352 i = nexti; 2353 /* printf(" pivot row i=%d\n",i); */ 2354 } 2355 2356 /* CHECK FOR ZERO PIVOT AND SAVE DIAGONAL ELEMENT */ 2357 if (dk == 0.0){ 2358 SETERRQ(PETSC_ERR_MAT_LU_ZRPVT,0,"Zero pivot"); 2359 } 2360 2361 /* SAVE NONZERO ENTRIES IN K-TH ROW OF U ... */ 2362 ba[k] = 1.0/dk; 2363 jmin = bi[k]; jmax = bi[k+1]; 2364 if (jmin < jmax) { 2365 for (j=jmin; j<jmax; j++){ 2366 juj = bj[j]; ba[j] = rtmp[juj]; rtmp[juj] = 0.0; 2367 } 2368 2369 /* ... AND ADD K TO ROW LIST FOR FIRST NONZERO ENTRY IN K-TH ROW */ 2370 il[k] = jmin; 2371 i = bj[jmin]; 2372 jl[k] = jl[i]; jl[i] = k; 2373 } 2374 } 2375 2376 ierr = PetscFree(rtmp);CHKERRQ(ierr); 2377 ierr = PetscFree(il);CHKERRQ(ierr); 2378 ierr = PetscFree(jl);CHKERRQ(ierr); 2379 2380 ierr = ISRestoreIndices(ip,&rip);CHKERRQ(ierr); 2381 C->factor = FACTOR_LU; 2382 C->assembled = PETSC_TRUE; 2383 PLogFlops(b->mbs); 2384 PetscFunctionReturn(0); 2385 } 2386 2387 #undef __FUNC__ 2388 #define __FUNC__ "MatLUFactor_SeqSBAIJ" 2389 int MatLUFactor_SeqSBAIJ(Mat A,IS row,IS col,MatLUInfo *info) 2390 { 2391 Mat_SeqBAIJ *mat = (Mat_SeqBAIJ*)A->data; 2392 int ierr,refct; 2393 Mat C; 2394 PetscOps *Abops; 2395 MatOps Aops; 2396 2397 PetscFunctionBegin; 2398 ierr = MatLUFactorSymbolic(A,row,col,info,&C);CHKERRQ(ierr); 2399 ierr = MatLUFactorNumeric(A,&C);CHKERRQ(ierr); 2400 2401 /* free all the data structures from mat */ 2402 ierr = PetscFree(mat->a);CHKERRQ(ierr); 2403 if (!mat->singlemalloc) { 2404 ierr = PetscFree(mat->i);CHKERRQ(ierr); 2405 ierr = PetscFree(mat->j);CHKERRQ(ierr); 2406 } 2407 if (mat->diag) {ierr = PetscFree(mat->diag);CHKERRQ(ierr);} 2408 if (mat->ilen) {ierr = PetscFree(mat->ilen);CHKERRQ(ierr);} 2409 if (mat->imax) {ierr = PetscFree(mat->imax);CHKERRQ(ierr);} 2410 if (mat->solve_work) {ierr = PetscFree(mat->solve_work);CHKERRQ(ierr);} 2411 if (mat->mult_work) {ierr = PetscFree(mat->mult_work);CHKERRQ(ierr);} 2412 if (mat->icol) {ierr = ISDestroy(mat->icol);CHKERRQ(ierr);} 2413 ierr = PetscFree(mat);CHKERRQ(ierr); 2414 2415 ierr = MapDestroy(A->rmap);CHKERRQ(ierr); 2416 ierr = MapDestroy(A->cmap);CHKERRQ(ierr); 2417 2418 /* 2419 This is horrible,horrible code. We need to keep the 2420 A pointers for the bops and ops but copy everything 2421 else from C. 2422 */ 2423 Abops = A->bops; 2424 Aops = A->ops; 2425 refct = A->refct; 2426 ierr = PetscMemcpy(A,C,sizeof(struct _p_Mat));CHKERRQ(ierr); 2427 mat = (Mat_SeqBAIJ*)A->data; 2428 PLogObjectParent(A,mat->icol); 2429 2430 A->bops = Abops; 2431 A->ops = Aops; 2432 A->qlist = 0; 2433 A->refct = refct; 2434 /* copy over the type_name and name */ 2435 ierr = PetscStrallocpy(C->type_name,&A->type_name);CHKERRQ(ierr); 2436 ierr = PetscStrallocpy(C->name,&A->name);CHKERRQ(ierr); 2437 2438 PetscHeaderDestroy(C); 2439 PetscFunctionReturn(0); 2440 } 2441 2442 2443