9 rm -f pgnnames.txt tmp.mtbl* open.mtbl.part????.new 2>/dev/null
12 echo "$FILE" >> pgnnames.txt
15 TS_BEGIN=$( date +%s%3N )
17 for X in $( seq 0 $(( MAPPERS - 1 )) ); do
22 START=$( ./find-pgn-split-point.sh "$FILE" $X $MAPPERS )
23 END=$( ./find-pgn-split-point.sh "$FILE" $(( X + 1 )) $MAPPERS )
24 ~/nmu/pgn-extract/pgn-extract --startpos $START --endpos $END --startfilenum $PGNNUM -e -Wsessebin "$FILE" 2>/dev/null
25 PGNNUM=$(( PGNNUM + 1 ))
27 ) | ./binloader - tmp.mtbl.mapper$X $SHARDS 500000
32 TS_MAPDONE=$( date +%s%3N )
33 MAPOUTPUT=$( du -sch tmp.mtbl.mapper* | tail -n 1 | cut -f1 | sed 's/\([^0-9]\)$/ \1/' )
37 parallel -i -j $REDUCERS bash -c 'X={}; ./binmerger tmp.mtbl.mapper*.part$( printf %04d $X ).subshard* open.mtbl.part$( printf %04d $X ).new' -- $( seq 0 $(( SHARDS - 1 )) )
39 TS_REDUCEDONE=$( date +%s%3N )
40 REDUCEOUTPUT=$( du -sch open.mtbl.part*.new | tail -n 1 | cut -f1 | sed 's/\([^0-9]\)$/ \1/' )
42 for X in $( seq 0 $(( SHARDS - 1 )) ); do
43 mv open.mtbl.part$( printf %04d $X ).new open.mtbl.part$( printf %04d $X )
47 TS_END=$( date +%s%3N )
50 printf " Map time: %7d ms\n" $(( TS_MAPDONE - TS_BEGIN ))
51 printf " Reduce time: %7d ms\n" $(( TS_REDUCEDONE - TS_MAPDONE ))
52 printf " Total time: %7d ms\n" $(( TS_END - TS_BEGIN ))
54 printf " Map output: %7sB\n" "$MAPOUTPUT"
55 printf " Reduce output: %7sB\n" "$REDUCEOUTPUT"