concurrency - MPI Barrier and Printing -

this question has answer here:

• ordering output in mpi 4 answers

i working on simple mpi assignment row wise matrix multiplication.

i trying output matrix , reason order of prints out of order. 1 process designated print @ time, output if flushed, , mpi_barrier used. confused how prints being reordered.

void print_matrix(int id, int p, int pn, int n, double **row, double *shared_col_data, double *resdata){     int i,j, k;     for(i=0; i<p; i++){         for(k=0; k < pn; k++){              int row_pos=((i*pn)+k);              if(id==i){                 if( row_pos <n){                     printf("[row: %10d][id: %3d]\t",row_pos,id);                     printf("|");                      for(j=0; j<n; j++){                         printf("%.4f ",row[k][j]);                     }                      if(row_pos == (n/2)){                         printf("| x |");                     }                     else{                         printf("|   |");                     }                      printf("%.4f ",shared_col_data[row_pos]);                     fflush(stdout);                 }             }//end ur turn print              mpi_barrier (mpi_comm_world);             if(id==0){                 if(row_pos == (n/2)){                     printf("| = |");                 }                 else{                     printf("|   |");                 }                 printf("%.4f|\n",resdata[row_pos]);                 fflush(stdout);              }              mpi_barrier (mpi_comm_world);         }       }//end processor loop   } 

here sample output:

input size(n): 12 processors(p): 12 n per processors(p): 1 id:  3 sn: 3 en: 3 id:  0 sn: 0 en: 0 id:  2 sn: 2 en: 2 id:  4 sn: 4 en: 4 id:  11 sn: 11 en: 11 id:  7 sn: 7 en: 7 id:  8 sn: 8 en: 8 id:  6 sn: 6 en: 6 id:  10 sn: 10 en: 10 id:  1 sn: 1 en: 1 id:  9 sn: 9 en: 9 id:  5 sn: 5 en: 5 data generated column data shared multiplication done res gather [row:          0][id:   0]  |0.5974 0.7066 0.9131 0.1548 0.4382 0.5132 0.3729 0.5554 0.7832 0.7953 0.5202 0.6986 |   |0.8076 |   |2.4959| [row:          1][id:   1]  |0.4320 0.9492 0.2266 0.1211 0.3904 0.9614 0.2000 0.7380 0.4471 0.3622 0.9844 0.1921 |   |0.0051 |   |2.3152| [row:          2][id:   2]  |0.2821 0.6740 0.9673 0.6623 0.6922 0.9760 0.8697 0.0096 0.6827 0.9590 0.2399 0.1100 |   |0.2254 |   |2.8286| [row:          3][id:   3]  |0.5915 0.1042 0.7262 0.8395 0.9665 0.9716 0.2252 0.7184 0.6054 0.8336 0.5033 0.2620 |   |0.3670 |   |2.8024| |   |2.1632| [row:          4][id:   4]  |0.0821 0.3956 0.0252 0.9953 0.3822 0.4278 0.8978 0.7726 0.5235 0.2972 0.3229 0.4520 |   |0.1409  [row:          5][id:   5]  |0.5684 0.0840 0.5961 0.7087 0.1331 0.1426 0.1554 0.3976 0.2051 0.1481 0.9468 0.7025 |   |0.5302 |   |2.1380| | = |2.7801| [row:          6][id:   6]  |0.7347 0.9194 0.3374 0.9823 0.1040 0.3878 0.7086 0.3132 0.4359 0.8223 0.2545 0.8752 | x |0.9129  [row:          7][id:   7]  |0.0464 0.6857 0.7146 0.6858 0.3210 0.2477 0.5767 0.2342 0.1406 0.5467 0.4063 0.0733 |   |0.1262 |   |1.8125| [row:          8][id:   8]  |0.6413 0.1076 0.2843 0.3515 0.9252 0.0349 0.0830 0.5063 0.9232 0.9900 0.5849 0.5612 |   |0.0204 |   |2.1263| [row:          9][id:   9]  |0.5292 0.3410 0.8543 0.5942 0.5822 0.3245 0.1719 0.9346 0.7611 0.3722 0.9653 0.4368 |   |0.5167 |   |2.4469| |   |2.6545| [row:         10][id:  10]  |0.7641 0.6008 0.9687 0.9276 0.2462 0.2832 0.0131 0.0390 0.4860 0.6569 0.9390 0.7620 |   |0.7972 |   |2.0755| [row:         11][id:  11]  |0.7291 0.5953 0.6171 0.0396 0.5790 0.5262 0.5881 0.2631 0.9517 0.5118 0.0835 0.2115 |   |0.2310  

you can see of lines begin ends of other lines, code prints surround barrier.

i feel simple issue , tired see it

thanks help.

mpi specification not cover parallel i/o. purely implementation dependant.