Resolved comments:

- Added postgres version query in queries.json
- Added pg_prewarm and get_stats functions in run-dann and run-hnsw.
- Deleted custom-dataset run and config script.
- Fixed key issues in hnsw and pvs-sbq run scripts

run-scripts/queries.json

@@ -1,4 +1,8 @@
 [
+    {
+        "description": "Postgresql Version",
+        "query": "SELECT version();"
+    },
     {
         "description": "List of extensions installed on postgersql server",
         "query": "SELECT extname AS name, extversion AS version FROM pg_extension ORDER BY extname;"

run-scripts/run-dann.py

@@ -40,11 +40,62 @@ def setup_database(config):
         cursor = conn.cursor()
         cursor.execute("CREATE EXTENSION IF NOT EXISTS vector;")
         cursor.execute("CREATE EXTENSION IF NOT EXISTS pg_diskann;")
+        cursor.execute("CREATE EXTENSION IF NOT EXISTS pg_buffercache;")
+        cursor.execute("CREATE EXTENSION IF NOT EXISTS pg_prewarm;")
         conn.commit()
         conn.close()
     except Exception as e:
         print(f"Setup failed: {e}")
 
+def get_stats(config):
+    with open('queries.json', 'r') as file:
+        queries = json.load(file)
+    try:
+        conn = psycopg2.connect(
+            dbname=config['db-name'],
+            user=config['username'],
+            password=config['password'],
+            host=config['host']
+        )
+        cur = conn.cursor()
+        for item in queries:
+            query = item['query']
+            description = item['description']
+            print(f"\nRunning query: {description}")
+            try:
+                cur.execute(query)
+                rows = cur.fetchall()
+                headers = [desc[0] for desc in cur.description]
+                print(f"{' | '.join(headers)}")
+                for row in rows:
+                    print(f"{' | '.join(map(str, row))}")
+            except Exception as e:
+                print(f"Failed to run query: {e}")
+        conn.close()
+    except Exception as e:
+        print(f"Setup failed: {e}")
+    finally:
+        conn.close()
+
+def pre_warm(config):
+    print(f"Running pre warm for database:{config['db-name']}")
+    try:
+        conn = psycopg2.connect(
+                dbname=config['db-name'],
+                user=config['username'],
+                password=config['password'],
+                host=config['host'],
+        )
+        cursor = conn.cursor()
+        cursor.execute("SELECT pg_prewarm('public.pgvector_index') as block_loaded")
+        conn.commit()
+
+        result = cursor.fetchone()
+        print(f"Pre-warm blocks loaded: {result[0]}")
+        conn.close()
+    except Exception as e:
+        print(f"Failed to pre-warm the database: {e}")
+
 def teardown_database(config):
     # Optionally drop the database after the test
     pass
@@ -186,6 +237,9 @@ def run_benchmark(case, db_config, benchmark_description):
                         current_configs = query_configurations(db_config)
                         for key, value in current_configs.items():
                             print(f"{key}: {value}")
+                        get_stats(db_config)
+                        f.flush()
+                        pre_warm(db_config)
                         print(f"Running command: {' '.join(command)}")
                         f.flush()
 
@@ -197,6 +251,9 @@ def run_benchmark(case, db_config, benchmark_description):
                     execution_time = end_time - start_time
                     print(f"total_duration={execution_time}")
                     print("***********END***********")
+                    with redirect_stdout(f):
+                        get_stats(db_config)
+                        f.flush()
                     f.flush()
             except subprocess.CalledProcessError as e:
                 print(f"Benchmark failed: {e}")

run-scripts/run-hnsw-bq.py

@@ -179,12 +179,12 @@ def run_benchmark(case, db_config, benchmark_description):
     else:
         base_command.append("--skip-load")
 
-    if case.get("search_serial", True):
+    if case.get("search-serial", True):
         base_command.append("--search-serial")
     else:
         base_command.append("--skip-search-serial")
 
-    if case.get("search_concurrent", True):
+    if case.get("search-concurrent", True):
         base_command.append("--search-concurrent")
     else:
         base_command.append("--skip-search-concurrent")