Kouhei Sutou
null+****@clear*****
Sun Nov 6 12:01:55 JST 2016
Kouhei Sutou 2016-11-06 12:01:55 +0900 (Sun, 06 Nov 2016) New Revision: 22a1fd6c27d6f62ba1152144c191e7e19e3e73ae https://github.com/groonga/wikipedia-search/commit/22a1fd6c27d6f62ba1152144c191e7e19e3e73ae Message: Run SELECTs in the same session Modified files: benchmark/centos7/en-postgresql.sh Modified: benchmark/centos7/en-postgresql.sh (+27 -20) =================================================================== --- benchmark/centos7/en-postgresql.sh 2016-11-05 18:59:44 +0900 (be1f61e) +++ benchmark/centos7/en-postgresql.sh 2016-11-06 12:01:55 +0900 (21e28c9) @@ -282,19 +282,20 @@ benchmark_search_pgroonga() { work_mem_size='64MB' work_mem="SET work_mem = '${work_mem_size}';" - search_path="SET search_path = \"\${user}\", public, pgroonga, pg_catalog;" enable_seqscan="SET enable_seqscan = no;" cat "${benchmark_dir}/en-search-words.list" | while read search_word; do + commands=() + commands=("--command" "${work_mem}" "${commands[@]}") + commands=("--command" "${enable_seqscan}" "${commands[@]}") + commands=("--command" "\\timing" "${commands[@]}") for i in $(seq ${n_search_tries}); do where="text @@ '${search_word}'" - echo "PGroonga: search: work_mem(${work_mem_size}): ${where}: ${i}:" - run sudo -u postgres -H psql -d ${pgroonga_db} \ - --command "${work_mem}" \ - --command "${search_path}" \ - --command "${enable_seqscan}" \ - --command "\\timing" \ - --command "SELECT COUNT(*) FROM wikipedia WHERE ${where}" + commands=("--command" + "SELECT COUNT(*) FROM wikipedia WHERE ${where}" + "${commands[@]}") done + echo "PGroonga: search: work_mem(${work_mem_size}): ${where}:" + run sudo -u postgres -H psql -d ${pgroonga_db} "${commands[@]}" done } @@ -304,16 +305,19 @@ benchmark_search_pg_trgm() work_mem="SET work_mem = '${work_mem_size}';" enable_seqscan="SET enable_seqscan = no;" cat "${benchmark_dir}/en-search-words.list" | while read search_word; do + commands=() + commands=("--command" "${work_mem}" "${commands[@]}") + commands=("--command" "${enable_seqscan}" "${commands[@]}") + commands=("--command" "\\timing" "${commands[@]}") for i in $(seq ${n_search_tries}); do where="text LIKE '%${search_word}%'" where=$(echo $where | sed -e "s/ OR /%' OR text LIKE '%/g") - echo "pg_trgm: search: work_mem(${work_mem_size}): ${where}: ${i}:" - run sudo -u postgres -H psql -d ${pg_trgm_db} \ - --command "${work_mem}" \ - --command "${enable_seqscan}" \ - --command "\\timing" \ - --command "SELECT COUNT(*) FROM wikipedia WHERE ${where}" + commands=("--command" + "SELECT COUNT(*) FROM wikipedia WHERE ${where}" + "${commands[@]}") done + echo "pg_trgm: search: work_mem(${work_mem_size}): ${where}" + run sudo -u postgres -H psql -d ${pg_trgm_db} "${commands[@]}" done } @@ -323,17 +327,20 @@ benchmark_search_textsearch() work_mem="SET work_mem = '${work_mem_size}';" enable_seqscan="SET enable_seqscan = no;" cat "${benchmark_dir}/en-search-words.list" | while read search_word; do + commands=() + commands=("--command" "${work_mem}" "${commands[@]}") + commands=("--command" "${enable_seqscan}" "${commands[@]}") + commands=("--command" "\\timing" "${commands[@]}") for i in $(seq ${n_search_tries}); do target="to_tsvector('english', text)" query="to_tsquery('english', '$(echo ${search_word} | sed -e 's/ OR / | /g')')" where="${target} @@ ${query}" - echo "textsearch: search: work_mem(${work_mem_size}): ${where}: ${i}:" - run sudo -u postgres -H psql -d ${textsearch_db} \ - --command "${work_mem}" \ - --command "${enable_seqscan}" \ - --command "\\timing" \ - --command "SELECT COUNT(*) FROM wikipedia WHERE ${where}" + commands=("--command" + "SELECT COUNT(*) FROM wikipedia WHERE ${where}" + "${commands[@]}") done + echo "textsearch: search: work_mem(${work_mem_size}): ${where}" + run sudo -u postgres -H psql -d ${textsearch_db} "${commands[@]}" done } -------------- next part -------------- HTML����������������������������...ダウンロード
