Updating benchmark.py

Also adding some logic to the QPE files to provide the datafile as an argument

Author: Jairik

QPEMPI

@@ -151,6 +151,12 @@ int main(int argc, char *argv[]) {
 
     set_rank(rank);
 
+    // Determine data file from CLI args or default
+    const char *dataFile = DATA_FILE;
+    if (argc > 1) {
+        dataFile = argv[1];
+    }
+
     // Start a timer for total runtime statistics
     double totalStart = MPI_Wtime();
 
@@ -159,7 +165,7 @@ int main(int argc, char *argv[]) {
         numOptimalIndexes,  // Number of indexes
         optimalIndexes,  // Indexes to build B+ trees for
         (const int *)optimalIndexTypes,  // Index types
-        DATA_FILE,
+        dataFile,
         TABLE_NAME
     );
 

QPEMPI.c

@@ -145,9 +145,22 @@ int main(int argc, char *argv[]) {
     printf("Starting main...\n"); fflush(stdout);
 
     // Pull out number of defined threads from CLI args (default to 8)
+    // Arg 1 can be Data File or Thread Count
     int num_threads = 8;
-    if (argc >= 2) {
-        num_threads = atoi(argv[1]);
+    const char *dataFile = DATA_FILE;
+
+    if (argc > 1) {
+        // If first char is digit, assume it's thread count (legacy support)
+        if (isdigit(argv[1][0])) {
+            num_threads = atoi(argv[1]);
+        } else {
+            // Otherwise it's the data file
+            dataFile = argv[1];
+            // And second arg is threads
+            if (argc > 2) {
+                num_threads = atoi(argv[2]);
+            }
+        }
     }
     omp_set_num_threads(num_threads);
 
@@ -160,7 +173,7 @@ int main(int argc, char *argv[]) {
         numOptimalIndexes,  // Number of indexes
         optimalIndexes,  // Indexes to build B+ trees for
         (const int *)optimalIndexTypes,  // Index types
-        DATA_FILE,
+        dataFile,
         TABLE_NAME
     );
     printf("Engine Initialized.\n"); fflush(stdout);

QPEOMP.c

@@ -15,8 +15,11 @@
 
 int main(int argc, char *argv[]) {
 
-    // NOTE: defaulting to sample queries file for ease of testing. Can implement CLI arg later.
-    (void)argc; (void)argv; // Unused for now. Prevent warnings.
+    // Determine data file from CLI args or default
+    const char *dataFile = DATA_FILE;
+    if (argc > 1) {
+        dataFile = argv[1];
+    }
 
     // Start a timer for total runtime statistics
     clock_t totalStart = clock();
@@ -26,7 +29,7 @@ int main(int argc, char *argv[]) {
         numOptimalIndexes,  // Number of indexes
         optimalIndexes,  // Indexes to build B+ trees for
         (const int *)optimalIndexTypes,  // Index types
-        DATA_FILE,
+        dataFile,
         TABLE_NAME
     );
 

QPESeq.c

@@ -15,13 +15,11 @@ def run_command(command, silent=False):
         if not silent:
             print(f"Error running command: {command}")
             print(f"Exit code: {e.returncode}")
-        # We don't raise here to allow the script to continue to cleanup
 
-def draw_menu(stdscr, selected_row_idx, options):
+def draw_menu(stdscr, selected_row_idx, options, title):
     stdscr.clear()
     h, w = stdscr.getmaxyx()
 
-    title = "Select Benchmark to Run (Use Arrow Keys + Enter)"
     stdscr.addstr(h//2 - len(options)//2 - 2, w//2 - len(title)//2, title)
 
     for idx, row in enumerate(options):
@@ -36,17 +34,10 @@ def draw_menu(stdscr, selected_row_idx, options):
 
     stdscr.refresh()
 
-def main(stdscr):
-    # Setup colors
-    curses.start_color()
-    curses.init_pair(1, curses.COLOR_BLACK, curses.COLOR_WHITE)
-    curses.curs_set(0)
-
-    options = ["Serial", "OMP", "MPI", "ALL", "Exit"]
+def select_option(stdscr, title, options):
     selected_row_idx = 0
-
     while True:
-        draw_menu(stdscr, selected_row_idx, options)
+        draw_menu(stdscr, selected_row_idx, options, title)
         key = stdscr.getch()
 
         if key == curses.KEY_UP and selected_row_idx > 0:
@@ -56,47 +47,82 @@ def main(stdscr):
         elif key == curses.KEY_ENTER or key in [10, 13]:
             return options[selected_row_idx]
 
+def get_datasets():
+    data_dir = "data-generation"
+    if not os.path.exists(data_dir):
+        return []
+    files = [f for f in os.listdir(data_dir) if f.endswith(".csv")]
+    files.sort()
+    # Return relative paths
+    return [os.path.join(data_dir, f) for f in files]
+
+def main(stdscr):
+    # Setup colors
+    curses.start_color()
+    curses.init_pair(1, curses.COLOR_BLACK, curses.COLOR_WHITE)
+    curses.curs_set(0)
+
+    # Select Dataset
+    datasets = get_datasets()
+    if not datasets:
+        return None, "No CSV files found in data-generation/"
+    
+    datasets.append("Exit")
+    dataset = select_option(stdscr, "Select Dataset (Use Arrow Keys + Enter)", datasets)
+    
+    if dataset == "Exit":
+        return "Exit", None
+
+    # Select Benchmark Version
+    benchmarks = ["Serial", "OMP", "MPI", "ALL", "Exit"]
+    benchmark = select_option(stdscr, "Select Benchmark to Run", benchmarks)
+
+    return benchmark, dataset
+
 def run_benchmark():
     # Run make silently
     print("Building project... (this may take a moment)")
     run_command("make", silent=True)
 
     # Show Menu
     try:
-        # curses.wrapper handles initialization and cleanup of curses
-        selection = curses.wrapper(main)
+        selection, dataset = curses.wrapper(main)
     except Exception as e:
         print(f"Error in UI: {e}")
-        # Ensure cleanup happens even if UI fails
         run_command("make clean", silent=True)
         return
 
-    if selection == "Exit":
+    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)
         return
 
-    print(f"\nRunning {selection} Benchmark...\n" + "="*40 + "\n")
+    print(f"\nRunning {selection} Benchmark with dataset: {dataset}\n" + "="*60 + "\n")
 
     # Run Selected
-    start_time = time.time()
+    start_time = time.time() 
+    
+    # Pass dataset as ARGS
+    args = f'ARGS="{dataset}"'
 
     if selection == "Serial":
-        run_command("make run")
+        run_command(f"make run {args}")
     elif selection == "OMP":
-        run_command("make run-omp")
+        run_command(f"make run-omp {args}")
     elif selection == "MPI":
-        run_command("make run-mpi")
+        run_command(f"make run-mpi {args}")
     elif selection == "ALL":
         print("--- Running Serial ---")
-        run_command("make run")
+        run_command(f"make run {args}")
         print("\n--- Running OMP ---")
-        run_command("make run-omp")
+        run_command(f"make run-omp {args}")
         print("\n--- Running MPI ---")
-        run_command("make run-mpi")
+        run_command(f"make run-mpi {args}")
 
     end_time = time.time()
-    print("\n" + "="*40)
+    print("\n" + "="*60)
     print(f"Total Benchmark Time: {end_time - start_time:.4f} seconds")
 
     # Run make clean silently

benchmark.py

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9881bfe6e9cbaaac36cc1c7b28b624a471d724d532449808fef26430df32b1f4
-size 125434931
+oid sha256:4cc6c639d9d2bb85214cc75f365516e1b08e01b8f66f45e284372b9902467313
+size 119604509

data-generation/commands_1m.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b303b920c6c6cd992eacb27f5c63110cc0fe4ede54609e453446f34ab1c62047
-size 5922674
+oid sha256:6e19fc03d3fc84a36555e3609127174c3ce1d598b6970f704f8e617244f2b539
+size 3445023

data-generation/commands_50k.csv

@@ -108,24 +108,24 @@ test: $(TEST_BINS)
 
 # Run the serial version
 run: QPESeq
-	./QPESeq
+	./QPESeq $(ARGS)
 
 # Default OpenMP thread count unless overridden
 OMP_THREADS ?= 4
 
 # Run OpenMP version correctly — env var applies only to this command
 run-omp: QPEOMP
 	@echo "Running with OMP_NUM_THREADS=$(OMP_THREADS)"
-	@env OMP_NUM_THREADS=$(OMP_THREADS) ./QPEOMP
+	@env OMP_NUM_THREADS=$(OMP_THREADS) ./QPEOMP $(ARGS)
 
 # Default MPI process count unless overridden
 MPI_PROCS ?= 4
 
 # Run MPI version with proper launcher syntax
 run-mpi: QPEMPI
 	@echo "Running with $(MPI_PROCS) MPI processes..."
-	@mpirun -np $(MPI_PROCS) ./QPEMPI 2>/dev/null \
-	 || mpiexec -np $(MPI_PROCS) ./QPEMPI
+	@mpirun -np $(MPI_PROCS) ./QPEMPI $(ARGS) 2>/dev/null \
+	 || mpiexec -np $(MPI_PROCS) ./QPEMPI $(ARGS)
 
 # Run QPESeq under Valgrind to check for memory leaks
 valgrind: QPESeq