GitLab

Branch codepoint-properties at commit #f203c6e2 has a modified xch which generates output to a ReturnedBuffer() rather than stdout.

However, this seems to be a performance bottleneck with excessive synchronization time required in p2s:

cameron@cs-osl-08:~/parabix-devel/build12$ sudo perf stat bin/xch -prop=slc ~/Wikibooks/wiki-books-all.xml -EnableCycleCounter >wislc1
CYCLE COUNTER:

  # NAME                                                                  ITEMS      CYCLES RATE  SYNC  PART  EXPD  COPY  PIPE  EXEC     %
  1 mmap_source16384@8                                                649901238     9971611  0.0  41.6   0.0   0.0   0.0   0.0  27.3   0.1 +-  0.0
  2 s2p8                                                              649901238   546614405  0.8   0.3   0.0   0.0   7.5   0.0  90.9   3.4 +-  0.0
  3 cc_7f8e068d851d690f52bf8d67c46df1bb0ffdeee1+CMCompressed          649901238    87582778  0.1   2.7   0.0   0.1   8.4   0.0  84.3   0.6 +-  0.0
  4 adjust_bixnum1x1+CMCompressed                                     649901238   183158518  0.3   1.1   0.1   0.0   0.0   0.0  96.6   1.2 +-  0.0
  5 unitInsertionExtractionMasks1x1_Before                            649901238    28898199  0.0   6.0   0.0   0.0   0.0   0.0  85.8   0.2 +-  0.0
  6 PopCountP8192                                                    1299802476    20029050  0.0   3.0   0.0   0.0   0.0   0.0  85.9   0.1 +-  0.0
  7 FilterByMask64__select_<i1>[1]@0:0_:_select_<i1>[1]@0:0_         1299802476   239533649  0.2   1.9   0.0   0.0   3.7   0.0  92.8   1.5 +-  0.0
  8 PopCountP2048                                                     649901242    18436352  0.0   5.0   0.3   0.0   0.1   0.0  73.8   0.1 +-  0.0
  9 streamExpand4:64_8:8                                              649901242   173845768  0.3   0.8   0.0   0.0   0.5   0.0  96.2   1.1 +-  0.0
 10 FieldDeposit64_8                                                  649901242   302462530  0.5   2.1   0.0   0.0  18.5   0.0  77.8   1.9 +-  0.0
 11 cc_8c7de60ab46889c81d5f889be46cb35d44eb1bb0+CMCompressed          649901242   153222112  0.2   3.2   0.0   0.0   6.4   0.0  88.0   1.0 +-  0.0
 12 cc_d7a768472fc15029ed17c5abad56ff65b4dc4693+CMCompressed          649901242   651987912  1.0  14.2   0.0   0.0  16.6   0.0  68.4   4.1 +-  0.0
 13 UTF8_BytePosition+CMCompressed                                    649901242   201231964  0.3   0.6   0.0   0.0   4.7   0.0  91.9   1.3 +-  0.0
 14 u8_delmask2x1+CMCompressed                                        649901242   189572156  0.3   1.6   0.0   0.0   2.7   0.0  93.6   1.2 +-  0.0
 15 PopCountP2048                                                     649901242    18526422  0.0   5.9   0.0   0.0   2.1   0.0  76.9   0.1 +-  0.0
 16 UTF8_Target_Class3x1+ins+del+CMCompressed                         649901242   177515785  0.3   0.3   0.0   0.0   0.9   0.0  95.9   1.1 +-  0.0
 17 u8_transform_bits_16x1+ins+del+CMCompressed                       649901242   572807532  0.9   3.3   0.0   0.0   0.0   0.0  95.7   3.6 +-  0.0
 18 fieldCompress64__select_<i1>[1]@0:0_:_select_<i1>[8]@0:01234567_  649901242   231152280  0.4   0.4   0.0   0.0   0.0   0.0  97.7   1.5 +-  0.0
 19 streamCompress64_8                                                649901242   294243971  0.5   0.2   0.0   0.0   3.6   0.0  94.0   1.8 +-  0.0
 20 p2s_8                                                             649889520 11796413425 18.2  65.2   0.0  32.6   0.0   0.0   2.1  74.1 +-  0.1

                                                                                 TOTAL:  49.3   0.0  24.2   1.6   0.0  24.2  99.9
xlated buffer length: 649889520

 Performance counter stats for 'bin/xch -prop=slc /home/cameron/Wikibooks/wiki-books-all.xml -EnableCycleCounter':

          5,661.78 msec task-clock                       #    2.816 CPUs utilized             
                67      context-switches                 #   11.834 /sec                      
                 0      cpu-migrations                   #    0.000 /sec                      
           487,093      page-faults                      #   86.032 K/sec                     
    15,706,883,693      cycles                           #    2.774 GHz                       
    20,438,537,075      instructions                     #    1.30  insn per cycle            
     4,525,549,271      branches                         #  799.315 M/sec                     
         2,807,389      branch-misses                    #    0.06% of all branches           

       2.010657317 seconds time elapsed

       4.601109000 seconds user
       1.060255000 seconds sys