X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=build-book.sh;h=94f3a298ef6c655f7c296fe24d7b3f1b4da7c462;hb=04d675c5a4a867b2c15e2ef64d5179353fd1489e;hp=e6e8c09afa4def12b1b1a07cb01868b634d215c7;hpb=e9b5b1e5b285542f44f0c52ee0b470b22ac7381b;p=remoteglot-book diff --git a/build-book.sh b/build-book.sh index e6e8c09..94f3a29 100755 --- a/build-book.sh +++ b/build-book.sh @@ -2,29 +2,60 @@ set -e +export MAPPERS=40 +export REDUCERS=40 export SHARDS=40 -export PARALLEL_LOADS=20 # Reduce if you have problems with OOM -rm -f pgnnames.txt part-*.bin part-*.mtbl part-*.mtbl.part???? open.mtbl.new open.mtbl.part???? open.mtbl.part????.new 2>/dev/null +rm -f pgnnames.txt tmp.mtbl* open.mtbl.part????.new 2>/dev/null -PGNNUM=0 for FILE in "$@"; do - date | tr -d "\n" - echo " $FILE" - ./parallel-parse-pgn.sh "$FILE" "$PGNNUM" echo "$FILE" >> pgnnames.txt - PGNNUM=$(( PGNNUM + 1 )) done -date -echo "Loading..." -parallel -i -j ${PARALLEL_LOADS} bash -c 'FILE="{}"; set -x; ./binloader "$FILE" "${FILE/bin/mtbl}" $SHARDS' -- part-*.bin +TS_BEGIN=$( date +%s%3N ) + +for X in $( seq 0 $(( MAPPERS - 1 )) ); do + ( + ( + PGNNUM=0 + for FILE in "$@"; do + COMP=0 + if echo "$FILE" | grep -q ^comp:; then + COMP=1 + FILE=$( echo "$FILE" | sed s/comp:// ) + fi + START=$( ./find-pgn-split-point.sh "$FILE" $X $MAPPERS ) + END=$( ./find-pgn-split-point.sh "$FILE" $(( X + 1 )) $MAPPERS ) + ~/nmu/pgn-extract/pgn-extract --computerflag $COMP --startpos $START --endpos $END --startfilenum $PGNNUM -e -Wsessebin "$FILE" 2>/dev/null + PGNNUM=$(( PGNNUM + 1 )) + done + ) | ./binloader - tmp.mtbl.mapper$X $SHARDS 500000 + ) & +done +wait + +TS_MAPDONE=$( date +%s%3N ) +MAPOUTPUT=$( du -sch tmp.mtbl.mapper* | tail -n 1 | cut -f1 | sed 's/\([^0-9]\)$/ \1/' ) + +echo "Reducing..." -rm -f part-*.bin +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 )) ) -parallel -i -j ${PARALLEL_LOADS} bash -c 'X={}; set -x; ./binmerger part-*.mtbl.part$( printf %04d $X ) open.mtbl.part$( printf %04d $X ).new' -- $( seq 0 $(( SHARDS - 1 )) ) +TS_REDUCEDONE=$( date +%s%3N ) +REDUCEOUTPUT=$( du -sch open.mtbl.part*.new | tail -n 1 | cut -f1 | sed 's/\([^0-9]\)$/ \1/' ) for X in $( seq 0 $(( SHARDS - 1 )) ); do mv open.mtbl.part$( printf %04d $X ).new open.mtbl.part$( printf %04d $X ) done -rm -f part-*.mtbl.part???? +rm -f tmp.mtbl* + +TS_END=$( date +%s%3N ) + +printf "\n" +printf " Map time: %7d ms\n" $(( TS_MAPDONE - TS_BEGIN )) +printf " Reduce time: %7d ms\n" $(( TS_REDUCEDONE - TS_MAPDONE )) +printf " Total time: %7d ms\n" $(( TS_END - TS_BEGIN )) +printf "\n" +printf " Map output: %7sB\n" "$MAPOUTPUT" +printf " Reduce output: %7sB\n" "$REDUCEOUTPUT" +printf "\n"