Merge pull request #472 from ClickHouse/various-tunes-2

alexey-milovidov · web-flow · commit 656ba67123a2 · 2025-07-05T08:42:09.000+03:00
Various tunes (2)
diff --git a/aurora-postgresql/README.md b/aurora-postgresql/README.md
@@ -41,7 +41,8 @@ gzip -d -f hits.tsv.gz
 
 psql -U postgres -h "${FQDN}" -t -c 'CREATE DATABASE test'
 psql -U postgres -h "${FQDN}" test -t < create.sql
-psql -U postgres -h "${FQDN}" test -t -c '\timing' -c "\\copy hits FROM 'hits.tsv'"
+echo -n "Load time: "
+command time -f '%e' psql -U postgres -h "${FQDN}" test -t -c "\\copy hits FROM 'hits.tsv'"
 ```
 
 > COPY 99997497
diff --git a/cedardb/benchmark.sh b/cedardb/benchmark.sh
@@ -26,11 +26,12 @@ until pg_isready -h localhost --dbname postgres -U postgres > /dev/null 2>&1; do
 # create table and ingest data
 PGPASSWORD=test  psql -h localhost -U postgres -t < create.sql
 echo "Inserting data..."
-PGPASSWORD=test time psql -h localhost -U postgres -t -c '\timing' -c "COPY hits FROM '/data/hits.tsv';"
+echo -n "Load time: "
+PGPASSWORD=test command time -f '%e' psql -h localhost -U postgres -t -c "COPY hits FROM '/data/hits.tsv';"
 
 # get ingested data size
-echo "data size after ingest:"
-PGPASSWORD=test psql -h localhost -U postgres -t -c '\timing' -c "SELECT pg_total_relation_size('hits');"
+echo -n "Data size: "
+PGPASSWORD=test psql -h localhost -U postgres -t -c "SELECT pg_total_relation_size('hits');"
 
 # run benchmark
 echo "running benchmark..."
diff --git a/citus/benchmark.sh b/citus/benchmark.sh
@@ -15,7 +15,8 @@ chmod 400 .pgpass
 
 psql -U postgres -h localhost -d postgres --no-password -t -c 'CREATE DATABASE test'
 psql -U postgres -h localhost -d postgres --no-password test -t < create.sql
-psql -U postgres -h localhost -d postgres --no-password test -t -c '\timing' -c "\\copy hits FROM 'hits.tsv'"
+echo -n "Load time: "
+command time -f '%e' psql -U postgres -h localhost -d postgres --no-password test -t -c "\\copy hits FROM 'hits.tsv'"
 
 # COPY 99997497
 # Time: 1579203.482 ms (26:19.203)
diff --git a/cratedb/benchmark.sh b/cratedb/benchmark.sh
@@ -40,7 +40,8 @@ chmod 444 /tmp/hits.tsv
 
 psql -U crate -h localhost --no-password -t < $CREATE_FILE
 
-psql -U crate -h localhost --no-password -t -c '\timing' -c "
+echo -n "Load time: "
+command time -f '%e' psql -U crate -h localhost --no-password -t -c "
   COPY hits
   FROM 'file:///tmp/hits.tsv'
   WITH
diff --git a/duckdb-memory/benchmark.sh b/duckdb-memory/benchmark.sh
@@ -21,3 +21,6 @@ cat log.txt | grep -P '^\d|Killed|Segmentation' | head -n1
 
 cat log.txt | grep -P '^\d|Killed|Segmentation' | tail -n+2 | sed -r -e 's/^.*(Killed|Segmentation).*$/null\nnull\nnull/' |
     awk '{ if (i % 3 == 0) { printf "[" }; printf $1; if (i % 3 != 2) { printf "," } else { print "]," }; ++i; }'
+
+echo -n "Data size: "
+grep -F 'Maximum resident set size' log.txt | grep -o -P '\d+$' | awk '{ print $1 * 1024 }'
diff --git a/hydra/benchmark.sh b/hydra/benchmark.sh
@@ -7,7 +7,8 @@ IFS=$'\n\t'
 #sudo apt-get install -y postgresql-client
 
 # load data
-psql "$DATABASE_URL" -c '\timing on' -t -f create.sql
+echo -n "Load time: "
+command time -f '%e' psql "$DATABASE_URL" -t -f create.sql
 
 # run test
 DATABASE_URL="$DATABASE_URL" ./run.sh 2>&1 | tee log.txt
diff --git a/infobright/benchmark.sh b/infobright/benchmark.sh
@@ -32,7 +32,7 @@ sudo docker exec mysql_ib du -bcs /mnt/mysql_data/ /usr/local/infobright-4.0.7-x
 
 # 13 760 341 294
 
-./run.sh 2>&1 | log
+./run.sh 2>&1 | tee log.txt
 
 cat log.txt |
   grep -P 'rows? in set|Empty set|^ERROR' |
diff --git a/oxla/benchmark.sh b/oxla/benchmark.sh
@@ -27,11 +27,12 @@ export PGCLIENTENCODING=UTF8
 
 PGPASSWORD=oxla psql -h localhost -U oxla -t < create.sql
 echo "Insert data."
-PGPASSWORD=oxla psql -h localhost -U oxla -t -c '\timing' -c "COPY hits FROM '/data/hits.csv';"
+echo -n "Load time: "
+PGPASSWORD=oxla command time -f '%e' psql -h localhost -U oxla -t -c "COPY hits FROM '/data/hits.csv';"
 
 # get ingested data size
-echo "data size after ingest:"
-PGPASSWORD=oxla psql -h localhost -U oxla -t -c '\timing' -c "SELECT pg_total_relation_size('hits');"
+echo -n "Data size: "
+PGPASSWORD=oxla psql -h localhost -U oxla -t -c "SELECT pg_total_relation_size('hits');"
 
 # run benchmark
 echo "running benchmark..."
diff --git a/pg_duckdb-motherduck/load.sh b/pg_duckdb-motherduck/load.sh
@@ -9,7 +9,5 @@ PARQUET_FILE='https://datasets.clickhouse.com/hits_compatible/hits.parquet'
 
 echo "Loading data"
 (
-    echo "\timing"
-    cat create.sql |
-        sed -e "s=REPLACE_SCHEMA=$DATABASE=g" -e "s=REPLACE_PARQUET_FILE=$PARQUET_FILE=g"
-) | psql --no-psqlrc --tuples-only $CONNECTION | grep 'Time'
+    cat create.sql | sed -e "s=REPLACE_SCHEMA=$DATABASE=g" -e "s=REPLACE_PARQUET_FILE=$PARQUET_FILE=g"
+) | psql --no-psqlrc --tuples-only $CONNECTION
diff --git a/pgpro_tam/benchmark.sh b/pgpro_tam/benchmark.sh
@@ -52,7 +52,8 @@ if [ "$1" == "parquet_fd_parall" ] ; then
     #insert data in parallel; not ordered insert is much faster, but breaks query performance
     sudo docker exec pgpro_tam bash -c "time cat /tmp/hits.tsv | parallel -l 2000000 -j 50 -N1 -k --spreadstdin 'psql -U postgres -t -c \"copy hits FROM STDIN\"'"
 else
-    psql -h 127.0.0.1 -U postgres -t -c '\timing' -c "COPY hits FROM '/tmp/hits.tsv'"
+    echo -n "Load time: "
+    command time -f '%e' psql -h 127.0.0.1 -U postgres -t -c "COPY hits FROM '/tmp/hits.tsv'"
 fi
 
 #run benchmark
diff --git a/tablespace/benchmark.sh b/tablespace/benchmark.sh
@@ -12,10 +12,14 @@ pigz -d -f hits.tsv.gz
 chmod 777 ~ hits.tsv
 
 psql "host=$HOSTNAME port=5432 dbname=csdb user=csuser password=$PASSWORD sslmode=require" < create.sql
-psql "host=$HOSTNAME port=5432 dbname=csdb user=csuser password=$PASSWORD sslmode=require" -t -c '\timing' -c "\\copy hits FROM 'hits.tsv'"
-psql "host=$HOSTNAME port=5432 dbname=csdb user=csuser password=$PASSWORD sslmode=require" < index.sql
-psql "host=$HOSTNAME port=5432 dbname=csdb user=csuser password=$PASSWORD sslmode=require" -t -c '\timing' -c "vacuum analyze hits"
-psql "host=$HOSTNAME port=5432 dbname=csdb user=csuser password=$PASSWORD sslmode=require" -t -c '\timing' -c "select columnstore.database_size('csdb')"
+echo -n "Load time: "
+command time -f '%e' psql "host=$HOSTNAME port=5432 dbname=csdb user=csuser password=$PASSWORD sslmode=require" -t -c "\\copy hits FROM 'hits.tsv'"
+echo -n "Load time: "
+command time -f '%e' psql "host=$HOSTNAME port=5432 dbname=csdb user=csuser password=$PASSWORD sslmode=require" < index.sql
+echo -n "Load time: "
+command time -f '%e' psql "host=$HOSTNAME port=5432 dbname=csdb user=csuser password=$PASSWORD sslmode=require" -t -c "vacuum analyze hits"
+echo -n 'Data size: '
+psql "host=$HOSTNAME port=5432 dbname=csdb user=csuser password=$PASSWORD sslmode=require" -t -c "select columnstore.database_size('csdb')"
 
 ./run.sh 2>&1 | tee log.txt
 
diff --git a/tembo-olap/benchmark.sh b/tembo-olap/benchmark.sh
@@ -13,9 +13,12 @@ chmod 777 ~ hits.tsv
 
 psql postgresql://postgres:$PASSWORD@$HOSTNAME:5432 -t -c 'CREATE DATABASE test'
 psql "host=$HOSTNAME port=5432 dbname=test user=postgres password=$PASSWORD sslmode=require" < create.sql
-psql "host=$HOSTNAME port=5432 dbname=test user=postgres password=$PASSWORD sslmode=require" -t -c '\timing' -c "\\copy hits FROM 'hits.tsv'"
-psql "host=$HOSTNAME port=5432 dbname=test user=postgres password=$PASSWORD sslmode=require" < index.sql
-psql "host=$HOSTNAME port=5432 dbname=test user=postgres password=$PASSWORD sslmode=require" -t -c '\timing' -c "select pg_total_relation_size('hits');"
+echo -n "Load time: "
+command time -f '%e' psql "host=$HOSTNAME port=5432 dbname=test user=postgres password=$PASSWORD sslmode=require" -t -c "\\copy hits FROM 'hits.tsv'"
+echo -n "Load time: "
+command time -f '%e' psql "host=$HOSTNAME port=5432 dbname=test user=postgres password=$PASSWORD sslmode=require" < index.sql
+echo -n "Data size: "
+psql "host=$HOSTNAME port=5432 dbname=test user=postgres password=$PASSWORD sslmode=require" -t -c "select pg_total_relation_size('hits');"
 
 ./run.sh "${HOSTNAME}" "${PASSWORD}" 2>&1 | tee log.txt
 
diff --git a/timescaledb-no-columnstore/benchmark.sh b/timescaledb-no-columnstore/benchmark.sh
@@ -29,10 +29,12 @@ sudo -u postgres psql nocolumnstore -c "CREATE INDEX ix_counterid ON hits (count
 sudo -u postgres psql -c "ALTER DATABASE nocolumnstore SET work_mem TO '1GB';"
 sudo -u postgres psql -c "ALTER DATABASE nocolumnstore SET min_parallel_table_scan_size TO '0';"
 
-sudo -u postgres psql nocolumnstore -t -c '\timing' -c "\\copy hits FROM 'hits.tsv'"
-sudo -u postgres psql nocolumnstore -t -c '\timing' -c "vacuum freeze analyze hits;"
+echo -n "Load time: "
+command time -f '%e' sudo -u postgres psql nocolumnstore -t -c "\\copy hits FROM 'hits.tsv'"
+echo -n "Load time: "
+command time -f '%e' sudo -u postgres psql nocolumnstore -t -c "vacuum freeze analyze hits;"
 
-#datasize
+echo -n "Data size: "
 sudo -u postgres psql nocolumnstore -c "\t" -c "SELECT hypertable_size('hits');"
 
 ./run.sh 2>&1 | tee log.txt
diff --git a/timescaledb/benchmark.sh b/timescaledb/benchmark.sh
@@ -31,16 +31,19 @@ sudo -u postgres psql -c "ALTER DATABASE test SET work_mem TO '1GB';"
 sudo -u postgres psql -c "ALTER DATABASE test SET min_parallel_table_scan_size TO '0';"
 sudo -u postgres psql test -c "SELECT enable_chunk_skipping('hits', 'counterid');"
 
-sudo -u postgres psql test -t -c '\timing' -c "\\copy hits FROM 'hits.tsv'"
+echo -n "Load time: "
+command time -f '%e' sudo -u postgres psql test -t -c "\\copy hits FROM 'hits.tsv'"
 
 # See https://github.com/timescale/timescaledb/issues/4473#issuecomment-1167095245
 # https://docs.timescale.com/timescaledb/latest/how-to-guides/compression/manually-compress-chunks/#compress-chunks-manually
 # TimescaleDB benchmark wihout compression is available in timescaledb no columnstore directory
 
-sudo -u postgres psql test -c "SELECT compress_chunk(i, if_not_compressed => true) FROM show_chunks('hits') i"
-sudo -u postgres psql test -t -c '\timing' -c "vacuum freeze analyze hits;"
+echo -n "Load time: "
+command time -f '%e' sudo -u postgres psql test -c "SELECT compress_chunk(i, if_not_compressed => true) FROM show_chunks('hits') i"
+echo -n "Load time: "
+command time -f '%e' sudo -u postgres psql test -t -c "vacuum freeze analyze hits;"
 
-#datasize
+echo -n "Data size: "
 sudo -u postgres psql test -c "\t" -c "SELECT hypertable_size('hits');"
 
 ./run.sh 2>&1 | tee log.txt
diff --git a/umbra/benchmark.sh b/umbra/benchmark.sh
@@ -28,7 +28,7 @@ sleep 5 # Things below fail otherwise ...
 start=$(date +%s%3N)
 PGPASSWORD=postgres psql -p 5432 -h 127.0.0.1 -U postgres -f create.sql
 end=$(date +%s%3N)
-echo "Load Time: $((end - start)) ms"
+echo "Load Time: $(( (end - start) / 1000 ))"
 
 ./run.sh 2>&1 | tee log.txt
 

Original file line number	Diff line number	Diff line change
`@@ -9,7 +9,5 @@ PARQUET_FILE='https://datasets.clickhouse.com/hits_compatible/hits.parquet'`
`9`	`9`
`10`	`10`	`echo "Loading data"`
`11`	`11`	`(`
`12`		`- echo "\timing"`
`13`		`- cat create.sql \|`
`14`		`- sed -e "s=REPLACE_SCHEMA=$DATABASE=g" -e "s=REPLACE_PARQUET_FILE=$PARQUET_FILE=g"`
`15`		`-) \| psql --no-psqlrc --tuples-only $CONNECTION \| grep 'Time'`
	`12`	`+ cat create.sql \| sed -e "s=REPLACE_SCHEMA=$DATABASE=g" -e "s=REPLACE_PARQUET_FILE=$PARQUET_FILE=g"`
	`13`	`+) \| psql --no-psqlrc --tuples-only $CONNECTION`