Add missing files for scaling factor 100

PingLiuPing · PingLiuPing · commit 3b883a0a8ba5 · 2025-04-24T20:29:16.000+01:00
diff --git a/benchmarks/tpch-load/cleanup_sf100.json b/benchmarks/tpch-load/cleanup_sf100.json
@@ -9,7 +9,7 @@
     "drop-table/partsupp.sql",
     "drop-table/region.sql",
     "drop-table/supplier.sql",
-    "drop-schema/drop_schema.sql"
+    "schema/drop_sf100.sql"
   ],
   "expected_row_counts": {
     "load_tpch_sf100_parquet": [
diff --git a/benchmarks/tpch-load/cleanup_sf1000.json b/benchmarks/tpch-load/cleanup_sf1000.json
@@ -0,0 +1,18 @@
+{
+  "description": "Cleanup tables.",
+  "query_files": [
+    "drop-table/customer.sql",
+    "drop-table/lineitem.sql",
+    "drop-table/nation.sql",
+    "drop-table/orders.sql",
+    "drop-table/part.sql",
+    "drop-table/partsupp.sql",
+    "drop-table/region.sql",
+    "drop-table/supplier.sql",
+    "schema/drop_sf1000.sql"
+  ],
+  "expected_row_counts": {
+    "load_tpch_sf1000_parquet": [
+    ]
+  }
+}
diff --git a/benchmarks/tpch-load/hive_sf100.json b/benchmarks/tpch-load/hive_sf100.json
@@ -1,5 +1,5 @@
 {
-  "description": "Specifies the schema for TPCH scale factor 10 GB.",
+  "description": "Specifies the schema for TPCH scale factor 100 GB.",
   "catalog": "hive",
   "schema": "load_tpch_sf100_parquet"
 }
diff --git a/benchmarks/tpch-load/hive_sf1000.json b/benchmarks/tpch-load/hive_sf1000.json
@@ -1,5 +1,5 @@
 {
-  "description": "Specifies the schema for TPCH scale factor 10 GB.",
+  "description": "Specifies the schema for TPCH scale factor 1000 GB.",
   "catalog": "hive",
   "schema": "load_tpch_sf1000_parquet"
 }
diff --git a/benchmarks/tpch-load/iceberg_sf100.json b/benchmarks/tpch-load/iceberg_sf100.json
@@ -1,5 +1,5 @@
 {
-  "description": "Specifies the schema for TPCH scale factor 10 GB.",
+  "description": "Specifies the schema for TPCH scale factor 100 GB.",
   "catalog": "iceberg",
   "schema": "load_tpch_sf100_parquet"
 }
diff --git a/benchmarks/tpch-load/iceberg_sf1000.json b/benchmarks/tpch-load/iceberg_sf1000.json
@@ -1,5 +1,5 @@
 {
-  "description": "Specifies the schema for TPCH scale factor 10 GB.",
+  "description": "Specifies the schema for TPCH scale factor 1000 GB.",
   "catalog": "iceberg",
   "schema": "load_tpch_sf1000_parquet"
 }
diff --git a/benchmarks/tpch-load/insert_sf1000_j.json b/benchmarks/tpch-load/insert_sf1000_j.json
@@ -1,5 +1,5 @@
 {
-  "description": "Loading tpch with scaling factor 10 to hive parquet tables.",
+  "description": "Loading tpch with scaling factor 1000 to hive/iceberg parquet tables.",
   "query_files": [
     "hive-tpch-parquet-j-1000/customer_sf1000.sql",
     "hive-tpch-parquet-j-1000/lineitem_sf1000.sql",
@@ -11,7 +11,7 @@
     "hive-tpch-parquet-j-1000/supplier_sf1000.sql"
   ],
   "expected_row_counts": {
-    "load_tpch_sf100_parquet": [
+    "load_tpch_sf1000_parquet": [
       1,
       1,
       1,
diff --git a/benchmarks/tpch-load/insert_sf1000_n.json b/benchmarks/tpch-load/insert_sf1000_n.json
@@ -1,5 +1,5 @@
 {
-  "description": "Loading tpch with scaling factor 10 to hive parquet tables.",
+  "description": "Loading tpch with scaling factor 1000 to hive/iceberg parquet tables.",
   "query_files": [
     "hive-tpch-parquet-n-1000/customer_sf1000.sql",
     "hive-tpch-parquet-n-1000/lineitem_sf1000.sql",
@@ -11,7 +11,7 @@
     "hive-tpch-parquet-n-1000/supplier_sf1000.sql"
   ],
   "expected_row_counts": {
-    "load_tpch_sf100_parquet": [
+    "load_tpch_sf1000_parquet": [
       1,
       1,
       1,
diff --git a/benchmarks/tpch-load/insert_sf100_j.json b/benchmarks/tpch-load/insert_sf100_j.json
@@ -1,5 +1,5 @@
 {
-  "description": "Loading tpch with scaling factor 10 to hive parquet tables.",
+  "description": "Loading tpch with scaling factor 100 to hive/iceberg parquet tables.",
   "query_files": [
     "hive-tpch-parquet-j/customer_sf100.sql",
     "hive-tpch-parquet-j/lineitem_sf100.sql",
diff --git a/benchmarks/tpch-load/insert_sf100_n.json b/benchmarks/tpch-load/insert_sf100_n.json
@@ -1,5 +1,5 @@
 {
-  "description": "Loading tpch with scaling factor 10 to hive parquet tables.",
+  "description": "Loading tpch with scaling factor 100 to hive/iceberg parquet tables.",
   "query_files": [
     "hive-tpch-parquet-n/customer_sf100.sql",
     "hive-tpch-parquet-n/lineitem_sf100.sql",
diff --git a/benchmarks/tpch-load/prepare_sf100.json b/benchmarks/tpch-load/prepare_sf100.json
@@ -1,7 +1,7 @@
 {
-  "description": "Loading tpch with scaling factor 10 to hive parquet tables.",
+  "description": "Loading tpch with scaling factor 100 to hive/iceberg parquet format tables.",
   "query_files": [
-    "create-schema/sf100.sql",
+    "schema/create_sf100.sql",
     "create-table/customer.sql",
     "create-table/lineitem.sql",
     "create-table/nation.sql",
diff --git a/benchmarks/tpch-load/prepare_sf1000.json b/benchmarks/tpch-load/prepare_sf1000.json
@@ -0,0 +1,27 @@
+{
+  "description": "Loading tpch with scaling factor 100 to hive/iceberg parquet format tables.",
+  "query_files": [
+    "schema/create_sf1000.sql",
+    "create-table/customer.sql",
+    "create-table/lineitem.sql",
+    "create-table/nation.sql",
+    "create-table/orders.sql",
+    "create-table/part.sql",
+    "create-table/partsupp.sql",
+    "create-table/region.sql",
+    "create-table/supplier.sql"
+  ],
+  "expected_row_counts": {
+    "load_tpch_sf1000_parquet": [
+      1,
+      1,
+      1,
+      1,
+      1,
+      1,
+      1,
+      1,
+      1
+    ]
+  }
+}
diff --git a/benchmarks/tpch-load/schema/create_sf100.sql b/benchmarks/tpch-load/schema/create_sf100.sql
@@ -0,0 +1,3 @@
+CREATE SCHEMA IF NOT EXISTS load_tpch_sf100_parquet WITH (LOCATION = 's3a://presto-workload/hive_native_load_tpch_sf100_parquet/');
+--Do not need to explicitly speficy schema here
+USE load_tpch_sf100_parquet;
diff --git a/benchmarks/tpch-load/schema/create_sf1000.sql b/benchmarks/tpch-load/schema/create_sf1000.sql
diff --git a/benchmarks/tpch-load/schema/drop_sf100.sql b/benchmarks/tpch-load/schema/drop_sf100.sql
@@ -0,0 +1 @@
+drop schema load_tpch_sf100_parquet;
diff --git a/benchmarks/tpch-load/schema/drop_sf1000.sql b/benchmarks/tpch-load/schema/drop_sf1000.sql

Original file line number	Diff line number	Diff line change
`@@ -1,5 +1,5 @@`
`1`	`1`	`{`
`2`		`- "description": "Specifies the schema for TPCH scale factor 10 GB.",`
	`2`	`+ "description": "Specifies the schema for TPCH scale factor 100 GB.",`
`3`	`3`	`"catalog": "hive",`
`4`	`4`	`"schema": "load_tpch_sf100_parquet"`
`5`	`5`	`}`
Original file line number	Diff line number	Diff line change
`@@ -1,5 +1,5 @@`
`1`	`1`	`{`
`2`		`- "description": "Loading tpch with scaling factor 10 to hive parquet tables.",`
	`2`	`+ "description": "Loading tpch with scaling factor 100 to hive/iceberg parquet tables.",`
`3`	`3`	`"query_files": [`
`4`	`4`	`"hive-tpch-parquet-j/customer_sf100.sql",`
`5`	`5`	`"hive-tpch-parquet-j/lineitem_sf100.sql",`