should be it

This commit is contained in:
2025-10-24 19:21:19 -05:00
parent a4b23fc57c
commit f09560c7b1
14047 changed files with 3161551 additions and 1 deletions

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpcds/csv/ingest_customer_demographics.benchmark
# description: benchmark ingestion of customer_demographics
# group: [csv]
template benchmark/ingestion/in_memory_db/tpcds/tpcds_ingestion_in_memory.benchmark.in
table_name=customer_demographics
format=csv

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpcds/csv/ingest_inventory.benchmark
# description: benchmark ingestion of inventory
# group: [csv]
template benchmark/ingestion/in_memory_db/tpcds/tpcds_ingestion_in_memory.benchmark.in
table_name=inventory
format=csv

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpcds/csv/ingest_store_sales.benchmark
# description: benchmark ingestion of store_sales
# group: [csv]
template benchmark/ingestion/in_memory_db/tpcds/tpcds_ingestion_in_memory.benchmark.in
table_name=store_sales
format=csv

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpcds/native/ingest_customer_demographics.benchmark
# description: benchmark ingestion of customer_demographics
# group: [native]
template benchmark/ingestion/in_memory_db/tpcds/native/tpcds_ingest_native.benchmark.in
table_name=customer_demographics
format=native

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpcds/native/ingest_inventory.benchmark
# description: benchmark ingestion of inventory
# group: [native]
template benchmark/ingestion/in_memory_db/tpcds/native/tpcds_ingest_native.benchmark.in
table_name=inventory
format=native

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpcds/native/ingest_store_sales.benchmark
# description: benchmark ingestion of store_sales
# group: [native]
template benchmark/ingestion/in_memory_db/tpcds/native/tpcds_ingest_native.benchmark.in
table_name=store_sales
format=native

View File

@@ -0,0 +1,26 @@
call dsdgen(sf=1);
create view call_center_native as select * from call_center;
create view household_demographics_native as select * from household_demographics;
create view store_returns_native as select * from store_returns;
create view catalog_page_native as select * from catalog_page;
create view income_band_native as select * from income_band;
create view store_sales_native as select * from store_sales;
create view catalog_returns_native as select * from catalog_returns;
create view inventory_native as select * from inventory;
create view time_dim_native as select * from time_dim;
create view catalog_sales_native as select * from catalog_sales;
create view item_native as select * from item;
create view warehouse_native as select * from warehouse;
create view customer_native as select * from customer;
create view promotion_native as select * from promotion;
create view web_page_native as select * from web_page;
create view customer_address_native as select * from customer_address;
create view reason_native as select * from reason;
create view web_returns_native as select * from web_returns;
create view customer_demographics_native as select * from customer_demographics;
create view ship_mode_native as select * from ship_mode;
create view web_sales_native as select * from web_sales;
create view date_dim_native as select * from date_dim;
create view store_native as select * from store;
create view web_site_native as select * from web_site;

View File

@@ -0,0 +1,19 @@
# name: ${FILE_PATH}
# description: ${DESCRIPTION}
# group: [ingestion]
name Q${table_name}
group ingestion
subgroup tpcds
require parquet
require tpcds
load benchmark/ingestion/in_memory_db/tpcds/native/load_native.sql
run
create table ${table_name}_ingested as select * from ${table_name}_${format};
cleanup
DROP TABLE ${table_name}_ingested;

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpcds/parquet/ingest_customer_demographics.benchmark
# description: benchmark ingestion of customer_demographics
# group: [parquet]
template benchmark/ingestion/in_memory_db/tpcds/tpcds_ingestion_in_memory.benchmark.in
table_name=customer_demographics
format=parquet

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpcds/parquet/ingest_inventory.benchmark
# description: benchmark ingestion of inventory
# group: [parquet]
template benchmark/ingestion/in_memory_db/tpcds/tpcds_ingestion_in_memory.benchmark.in
table_name=inventory
format=parquet

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpcds/parquet/ingest_store_sales.benchmark
# description: benchmark ingestion of store_sales
# group: [parquet]
template benchmark/ingestion/in_memory_db/tpcds/tpcds_ingestion_in_memory.benchmark.in
table_name=store_sales
format=parquet

View File

@@ -0,0 +1,23 @@
# name: ${FILE_PATH}
# description: ${DESCRIPTION}
# group: [ingestion]
name Q${table_name}
group ingestion
subgroup tpcds
require tpcds
require parquet
cache tpcds_sf1_ingest.duckdb no_connect
load benchmark/ingestion/load_tpcds.sql
# reads table from view created in load script
run
create table ${table_name}_ingested as select * from read_${format}('${BENCHMARK_DIR}/tpcds_${format}/${table_name}.${format}');
cleanup
drop table ${table_name}_ingested;

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpch/csv/ingest_lineitem.benchmark
# description: benchmark ingestion of lineitem
# group: [csv]
template benchmark/ingestion/in_memory_db/tpch/tpch_ingestion_in_memory.benchmark.in
table_name=lineitem
format=csv

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpch/csv/ingest_orders.benchmark
# description: benchmark ingestion of orders
# group: [csv]
template benchmark/ingestion/in_memory_db/tpch/tpch_ingestion_in_memory.benchmark.in
table_name=orders
format=csv

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpch/native/ingest_lineitem.benchmark
# description: benchmark ingestion of lineitem
# group: [native]
template benchmark/ingestion/in_memory_db/tpch/native/tpch_ingest_native.benchmark.in
table_name=lineitem
format=native

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpch/native/ingest_orders.benchmark
# description: benchmark ingestion of orders
# group: [native]
template benchmark/ingestion/in_memory_db/tpch/native/tpch_ingest_native.benchmark.in
table_name=orders
format=native

View File

@@ -0,0 +1,10 @@
call dbgen(sf=1);
create view customer_native as select * from customer;
create view lineitem_native as select * from lineitem;
create view nation_native as select * from nation;
create view orders_native as select * from orders;
create view part_native as select * from part;
create view partsupp_native as select * from partsupp;
create view region_native as select * from region;
create view supplier_native as select * from supplier;

View File

@@ -0,0 +1,19 @@
# name: ${FILE_PATH}
# description: ${DESCRIPTION}
# group: [ingestion]
name Q${table_name}
group ingestion
subgroup tpch
require parquet
require tpch
load benchmark/ingestion/in_memory_db/tpch/native/load_native.sql
run
create table ${table_name}_ingested as select * from ${table_name}_${format};
cleanup
DROP TABLE ${table_name}_ingested;

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpch/parquet/ingest_lineitem.benchmark
# description: benchmark ingestion of lineitem
# group: [parquet]
template benchmark/ingestion/in_memory_db/tpch/tpch_ingestion_in_memory.benchmark.in
table_name=lineitem
format=parquet

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/in_memory_db/tpch/parquet/ingest_orders.benchmark
# description: benchmark ingestion of orders
# group: [parquet]
template benchmark/ingestion/in_memory_db/tpch/tpch_ingestion_in_memory.benchmark.in
table_name=orders
format=parquet

View File

@@ -0,0 +1,21 @@
# name: ${FILE_PATH}
# description: ${DESCRIPTION}
# group: [ingestion]
name Q${table_name}
group ingestion
subgroup tpch
require parquet
require tpch
cache tpch_sf1_ingest.duckdb no_connect
load benchmark/ingestion/load_tpch.sql
run
create table ${table_name}_ingested as select * from read_${format}('${BENCHMARK_DIR}/tpch_${format}/${table_name}.${format}');
cleanup
DROP TABLE ${table_name}_ingested;

View File

@@ -0,0 +1,99 @@
CALL dsdgen(sf=1);
EXPORT DATABASE 'duckdb_benchmark_data/tpcds_parquet' (FORMAT PARQUET);
EXPORT DATABASE 'duckdb_benchmark_data/tpcds_csv' (FORMAT CSV);
create view call_center_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/call_center.csv');
create view call_center_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/call_center.parquet');
create view call_center_native as select * from call_center;
create view household_demographics_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/household_demographics.csv');
create view household_demographics_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/household_demographics.parquet');
create view household_demographics_native as select * from household_demographics;
create view store_returns_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/store_returns.csv');
create view store_returns_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/store_returns.parquet');
create view store_returns_native as select * from store_returns;
create view catalog_page_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/catalog_page.csv');
create view catalog_page_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/catalog_page.parquet');
create view catalog_page_native as select * from catalog_page;
create view income_band_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/income_band.csv');
create view income_band_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/income_band.parquet');
create view income_band_native as select * from income_band;
create view store_sales_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/store_sales.csv');
create view store_sales_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/store_sales.parquet');
create view store_sales_native as select * from store_sales;
create view catalog_returns_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/catalog_returns.csv');
create view catalog_returns_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/catalog_returns.parquet');
create view catalog_returns_native as select * from catalog_returns;
create view inventory_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/inventory.csv');
create view inventory_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/inventory.parquet');
create view inventory_native as select * from inventory;
create view time_dim_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/time_dim.csv');
create view time_dim_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/time_dim.parquet');
create view time_dim_native as select * from time_dim;
create view catalog_sales_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/catalog_sales.csv');
create view catalog_sales_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/catalog_sales.parquet');
create view catalog_sales_native as select * from catalog_sales;
create view item_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/item.csv');
create view item_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/item.parquet');
create view item_native as select * from item;
create view warehouse_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/warehouse.csv');
create view warehouse_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/warehouse.parquet');
create view warehouse_native as select * from warehouse;
create view customer_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/customer.csv');
create view customer_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/customer.parquet');
create view customer_native as select * from customer;
create view promotion_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/promotion.csv');
create view promotion_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/promotion.parquet');
create view promotion_native as select * from promotion;
create view web_page_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/web_page.csv');
create view web_page_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/web_page.parquet');
create view web_page_native as select * from web_page;
create view customer_address_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/customer_address.csv');
create view customer_address_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/customer_address.parquet');
create view customer_address_native as select * from customer_address;
create view reason_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/reason.csv');
create view reason_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/reason.parquet');
create view reason_native as select * from reason;
create view web_returns_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/web_returns.csv');
create view web_returns_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/web_returns.parquet');
create view web_returns_native as select * from web_returns;
create view customer_demographics_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/customer_demographics.csv');
create view customer_demographics_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/customer_demographics.parquet');
create view customer_demographics_native as select * from customer_demographics;
create view ship_mode_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/ship_mode.csv');
create view ship_mode_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/ship_mode.parquet');
create view ship_mode_native as select * from ship_mode;
create view web_sales_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/web_sales.csv');
create view web_sales_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/web_sales.parquet');
create view web_sales_native as select * from web_sales;
create view date_dim_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/date_dim.csv');
create view date_dim_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/date_dim.parquet');
create view date_dim_native as select * from date_dim;
create view store_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/store.csv');
create view store_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/store.parquet');
create view store_native as select * from store;
create view web_site_csv as select * from read_csv('duckdb_benchmark_data/tpcds_csv/web_site.csv');
create view web_site_parquet as select * from read_parquet('duckdb_benchmark_data/tpcds_parquet/web_site.parquet');
create view web_site_native as select * from web_site;

View File

@@ -0,0 +1,35 @@
CALL dbgen(sf=1);
EXPORT DATABASE 'duckdb_benchmark_data/tpch_parquet' (FORMAT PARQUET);
EXPORT DATABASE 'duckdb_benchmark_data/tpch_csv' (FORMAT CSV);
create view customer_csv as select * from read_csv('duckdb_benchmark_data/tpch_csv/customer.csv');
create view customer_parquet as select * from read_parquet('duckdb_benchmark_data/tpch_parquet/customer.parquet');
create view customer_native as select * from customer;
create view lineitem_csv as select * from read_csv('duckdb_benchmark_data/tpch_csv/lineitem.csv');
create view lineitem_parquet as select * from read_parquet('duckdb_benchmark_data/tpch_parquet/lineitem.parquet');
create view lineitem_native as select * from lineitem;
create view nation_csv as select * from read_csv('duckdb_benchmark_data/tpch_csv/nation.csv');
create view nation_parquet as select * from read_parquet('duckdb_benchmark_data/tpch_parquet/nation.parquet');
create view nation_native as select * from nation;
create view orders_csv as select * from read_csv('duckdb_benchmark_data/tpch_csv/orders.csv');
create view orders_parquet as select * from read_parquet('duckdb_benchmark_data/tpch_parquet/orders.parquet');
create view orders_native as select * from orders;
create view part_csv as select * from read_csv('duckdb_benchmark_data/tpch_csv/part.csv');
create view part_parquet as select * from read_parquet('duckdb_benchmark_data/tpch_parquet/part.parquet');
create view part_native as select * from part;
create view partsupp_csv as select * from read_csv('duckdb_benchmark_data/tpch_csv/partsupp.csv');
create view partsupp_parquet as select * from read_parquet('duckdb_benchmark_data/tpch_parquet/partsupp.parquet');
create view partsupp_native as select * from partsupp;
create view region_csv as select * from read_csv('duckdb_benchmark_data/tpch_csv/region.csv');
create view region_parquet as select * from read_parquet('duckdb_benchmark_data/tpch_parquet/region.parquet');
create view region_native as select * from region;
create view supplier_csv as select * from read_csv('duckdb_benchmark_data/tpch_csv/supplier.csv');
create view supplier_parquet as select * from read_parquet('duckdb_benchmark_data/tpch_parquet/supplier.parquet');
create view supplier_native as select * from supplier;

View File

@@ -0,0 +1,8 @@
# name: benchmark/ingestion/persistent_storage/tpcds/csv/ingest_inventory.benchmark
# description: benchmark ingestion of inventory
# group: [csv]
template benchmark/ingestion/persistent_storage/tpcds/tpcds_ingestion_persistent.benchmark.in
table_name=inventory
format=csv

View File

@@ -0,0 +1,8 @@
# name: benchmark/ingestion/persistent_storage/tpcds/csv/ingest_store_sales.benchmark
# description: benchmark ingestion of store_sales
# group: [csv]
template benchmark/ingestion/persistent_storage/tpcds/tpcds_ingestion_persistent.benchmark.in
table_name=store_sales
format=csv

View File

@@ -0,0 +1,8 @@
# name: benchmark/ingestion/persistent_storage/tpcds/native/ingest_inventory.benchmark
# description: benchmark ingestion of inventory
# group: [native]
template benchmark/ingestion/persistent_storage/tpcds/tpcds_ingestion_persistent.benchmark.in
table_name=inventory
format=native

View File

@@ -0,0 +1,8 @@
# name: benchmark/ingestion/persistent_storage/tpcds/native/ingest_store_sales.benchmark
# description: benchmark ingestion of store_sales
# group: [native]
template benchmark/ingestion/persistent_storage/tpcds/tpcds_ingestion_persistent.benchmark.in
table_name=store_sales
format=native

View File

@@ -0,0 +1,8 @@
# name: benchmark/ingestion/persistent_storage/tpcds/parquet/ingest_inventory.benchmark
# description: benchmark ingestion of inventory
# group: [parquet]
template benchmark/ingestion/persistent_storage/tpcds/tpcds_ingestion_persistent.benchmark.in
table_name=inventory
format=parquet

View File

@@ -0,0 +1,8 @@
# name: benchmark/ingestion/persistent_storage/tpcds/parquet/ingest_store_sales.benchmark
# description: benchmark ingestion of store_sales
# group: [parquet]
template benchmark/ingestion/persistent_storage/tpcds/tpcds_ingestion_persistent.benchmark.in
table_name=store_sales
format=parquet

View File

@@ -0,0 +1,22 @@
# name: ${FILE_PATH}
# description: ${DESCRIPTION}
# group: [ingestion]
name Q${table_name}
group ingestion
subgroup tpcds
require tpcds
require parquet
cache tpcds_sf1_ingest.duckdb
load benchmark/ingestion/load_tpcds.sql
# reads table from view created in load script
run
create table ${table_name}_ingested as select * from ${table_name}_${format};
cleanup
drop table ${table_name}_ingested;

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/persistent_storage/tpch/csv/ingest_lineitem.benchmark
# description: benchmark ingestion of lineitem
# group: [csv]
template benchmark/ingestion/persistent_storage/tpch/tpch_ingestion_persistent.benchmark.in
table_name=lineitem
format=csv

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/persistent_storage/tpch/csv/ingest_orders.benchmark
# description: benchmark ingestion of orders
# group: [csv]
template benchmark/ingestion/persistent_storage/tpch/tpch_ingestion_persistent.benchmark.in
table_name=orders
format=csv

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/persistent_storage/tpch/native/ingest_lineitem.benchmark
# description: benchmark ingestion of lineitem
# group: [native]
template benchmark/ingestion/persistent_storage/tpch/tpch_ingestion_persistent.benchmark.in
table_name=lineitem
format=native

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/persistent_storage/tpch/native/ingest_orders.benchmark
# description: benchmark ingestion of orders
# group: [native]
template benchmark/ingestion/persistent_storage/tpch/tpch_ingestion_persistent.benchmark.in
table_name=orders
format=native

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/persistent_storage/tpch/parquet/ingest_lineitem.benchmark
# description: benchmark ingestion of lineitem
# group: [parquet]
template benchmark/ingestion/persistent_storage/tpch/tpch_ingestion_persistent.benchmark.in
table_name=lineitem
format=parquet

View File

@@ -0,0 +1,7 @@
# name: benchmark/ingestion/persistent_storage/tpch/parquet/ingest_orders.benchmark
# description: benchmark ingestion of orders
# group: [parquet]
template benchmark/ingestion/persistent_storage/tpch/tpch_ingestion_persistent.benchmark.in
table_name=orders
format=parquet

View File

@@ -0,0 +1,21 @@
# name: ${FILE_PATH}
# description: ${DESCRIPTION}
# group: [ingestion]
name Q${table_name}
group ingestion
subgroup tpch
require parquet
require tpch
cache tpch_sf1_ingest.duckdb
load benchmark/ingestion/load_tpch.sql
run
create table ${table_name}_ingested as select * from ${table_name}_${format};
cleanup
DROP TABLE ${table_name}_ingested;