Improve comparison.

result_compare/compare.py

@@ -1,30 +1,66 @@
 import numpy as np
+import hashlib
 import os
+from collections import defaultdict
 
 results_folder = "results"
-baseline_file = "baseline_dp_O3"
+baseline_file_sp = "baseline_sp_O0"
+baseline_file_dp = "baseline_dp_O0"
+
 
 def read_result(file_path):
     with open(file_path, 'r') as file:
         data = file.read()
         matrix = np.array([list(map(int, row.split(','))) for row in data.strip().split('\n')])
         return matrix
 
-baseline_matrix = read_result(os.path.join(results_folder, baseline_file))
-print(f"Baseline file is {baseline_file}")
+# Optimized function to compute the hash of a file using a larger buffer size
+def compute_file_hash(file_path, hash_algorithm='sha256', chunk_size=65536):
+    hash_func = hashlib.new(hash_algorithm)  # Choose hash algorithm
+
+    with open(file_path, 'rb') as file:
+        while chunk := file.read(chunk_size):  # Read in larger chunks (e.g., 64KB)
+            hash_func.update(chunk)
+
+    return hash_func.hexdigest()  # Return the hexadecimal representation of the hash
 
+file_hashes = defaultdict(list) 
 # Loop through all files in the results folder (excluding the baseline file)
 for filename in sorted(os.listdir(results_folder)):
-
+    file_hashes[compute_file_hash(os.path.join(results_folder, filename))].append(filename)
+
+
+baseline_sp = read_result(os.path.join(results_folder, baseline_file_sp))
+baseline_dp = read_result(os.path.join(results_folder, baseline_file_dp))
+sp_groups = 0
+dp_groups = 0
+
+baseline_difference = np.abs(baseline_dp - baseline_sp)
+baseline_difference_sum = np.sum(baseline_difference)
+baseline_difference_average = np.mean(baseline_difference)
+print(f"Difference baseline_dp baseline_sp: {baseline_difference_sum} {baseline_difference_average}")
+
+for i, hash_group in enumerate(sorted(file_hashes.items(), key=lambda x: len(x[1]))):
+    _hash, group = hash_group
     # Construct the full path of the current result file
-    file_path = os.path.join(results_folder, filename)
+    file_path = os.path.join(results_folder, group[0])
 
     # Read the current result file into a NumPy array
     result_matrix = read_result(file_path)
 
-    # Compare the result matrix with the baseline matrix
-    if np.array_equal(result_matrix, baseline_matrix):
-        print(f"File {filename} matches the baseline.")
-    #else:
-    #    print(f"File {filename} does NOT match the baseline.")
+    baseline_matrix = baseline_dp
+    if  "_sp_" in group[0]:
+        baseline_matrix = baseline_sp
+        sp_groups += 1
+    else:
+        dp_groups += 1
+
+    difference = np.abs(baseline_matrix - result_matrix)
+    difference_sum = np.sum(difference)
+    difference_average = np.mean(difference)
+    print(f"Group {_hash} {i}:")
+    print(f'\t{"\n\t".join(group)}')
+    print(f"\tDifference Sum: {difference_sum} {difference_average}")
+
 
+print(f"\nTotal groups: {len(file_hashes)} sp: {sp_groups} dp: {dp_groups}")