Benchmarking optimizations

Making this way faster by removing redundant builds and whatnot.

Author: Jairik

benchmark.py

@@ -4,7 +4,7 @@
 import os
 import time
 
-def run_command(command, silent=False):
+def run_command(command, silent=False, dataset=None):
     """Runs a shell command."""
     try:
         if silent:
@@ -89,14 +89,14 @@ def run_benchmark():
         selection, dataset = curses.wrapper(main)
     except Exception as e:
         print(f"Error in UI: {e}")
-        run_command("make clean", silent=True)
+        # run_command("make clean", silent=True)
         return
 
     if selection == "Exit" or selection is None:
         if dataset and dataset != "Exit":
              print(dataset) # Print error message if any
         print("Exiting...")
-        run_command("make clean", silent=True)
+        # run_command("make clean", silent=True)
         return
 
     print(f"\nRunning {selection} Benchmark with dataset: {dataset}\n" + "="*60 + "\n")
@@ -105,30 +105,36 @@ def run_benchmark():
     start_time = time.time() 
 
     # Pass dataset as ARGS
-    args = f'ARGS="{dataset}"'
+    # args = f'ARGS="{dataset}"'
 
+    # Run the executable of the select benchmark, avoid make rules for overhead concerns
     if selection == "Serial":
-        run_command(f"make run {args}")
+        run_command(f"./QPESeq {dataset}")
+    
     elif selection == "OMP":
-        run_command(f"make run-omp {args}")
+        run_command(f"./QPEOMP {dataset}")
+    
     elif selection == "MPI":
-        run_command(f"make run-mpi {args}")
+        run_command(f"./QPEMPI {dataset}")
+    
     elif selection == "ALL":
         print("--- Running Serial ---")
-        run_command(f"make run {args}")
+        run_command(f"./QPESeq {dataset}")
+        
         print("\n--- Running OMP ---")
-        run_command(f"make run-omp {args}")
+        run_command(f"./QPEOMP {dataset}")
+        
         print("\n--- Running MPI ---")
-        run_command(f"make run-mpi {args}")
+        run_command(f"./QPEMPI {dataset}")
 
     end_time = time.time()
     print("\n" + "="*60)
     print(f"Total Benchmark Time: {end_time - start_time:.4f} seconds")
 
-    # Run make clean silently
-    print("Cleaning up build artifacts...")
-    run_command("make clean", silent=True)
-    print("Done.")
+    # # Run make clean silently
+    # print("Cleaning up build artifacts...")
+    # run_command("make clean", silent=True)
+    # print("Done.")
 
 if __name__ == "__main__":
     run_benchmark()

data-generation/commands_1m.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4cc6c639d9d2bb85214cc75f365516e1b08e01b8f66f45e284372b9902467313
-size 119604509
+oid sha256:f971b4bd2192c287507ba438684a974d4b2b1b785b33119061126735ea9f0a83
+size 119604440

sample-queries-FULL.txt

@@ -0,0 +1,36 @@
+# -- SAMPLE SQL QUERIES --
+# -- Sample 1:
+SELECT command_id, base_command, sudo_used, user_name, timestamp
+FROM Commands
+WHERE sudo_used = FALSE AND user_name = "student1030";
+
+# -- Sample 2:
+SELECT command_id, raw_command, user_name, risk_level, timestamp
+FROM Commands
+WHERE sudo_used = TRUE AND risk_level > 2;
+
+# -- Sample 3:
+SELECT raw_command, exit_code, timestamp, sudo_used, user_name, risk_level
+FROM Commands
+WHERE risk_level > 3;
+
+# -- Sample 4:
+SELECT *
+FROM Commands
+WHERE risk_level = 5;
+
+# -- Sample 5:
+INSERT INTO Commands VALUES (999999, "echo 'test insert'", "echo", "bash", 0, "2025-12-01T12:00:00.000Z", "FALSE", "/home/test", 1000, "testuser", "test-host", 1);
+
+# -- Sample 6:
+DELETE FROM Commands WHERE command_id = 999999;
+
+# -- Sample 7:
+SELECT command_id, raw_command, risk_level, exit_code
+FROM Commands
+WHERE sudo_used = TRUE OR (risk_level = 5 AND shell_type = "bash");
+
+# -- Sample 8:
+SELECT user_name, working_directory, base_command
+FROM Commands
+WHERE user_id = 1001 OR (user_name = "student1002" AND shell_type = "zsh");

sample-queries.txt

@@ -22,9 +22,6 @@ WHERE risk_level = 5;
 # -- Sample 5:
 INSERT INTO Commands VALUES (999999, "echo 'test insert'", "echo", "bash", 0, "2025-12-01T12:00:00.000Z", "FALSE", "/home/test", 1000, "testuser", "test-host", 1);
 
-# -- Sample 6:
-DELETE FROM Commands WHERE command_id = 999999;
-
 # -- Sample 7:
 SELECT command_id, raw_command, risk_level, exit_code
 FROM Commands