diff --git a/chesspp/simulation.py b/chesspp/simulation.py
index 98cf94c..52c25f2 100644
--- a/chesspp/simulation.py
+++ b/chesspp/simulation.py
@@ -48,11 +48,15 @@ class Evaluation:
 
     def run(self, n_games=100, proc=mp.cpu_count()) -> List[EvaluationResult]:
         proc = min(proc, mp.cpu_count())
-        with mp.Pool(proc) as pool:
-            args = [(self.engine_a, self.strategy_a, self.engine_b, self.strategy_b, self.limit, self.stockfish_path, self.lc0_path) for i
-                    in
-                    range(n_games)]
-            return pool.map(Evaluation._test_simulate, args)
+        arg = (self.engine_a, self.strategy_a, self.engine_b, self.strategy_b, self.limit, self.stockfish_path, self.lc0_path)
+        if proc > 1:
+            with mp.Pool(proc) as pool:
+                args = [arg for i in range(n_games)]
+                return pool.map(Evaluation._test_simulate, args)
+        return [
+            Evaluation._test_simulate(arg)
+            for _ in range(n_games)
+        ]
 
     @staticmethod
     def _test_simulate(arg: Tuple[EngineEnum, StrategyEnum, EngineEnum, StrategyEnum, Limit, str, str]) -> EvaluationResult:
diff --git a/chesspp/util_gaussian.py b/chesspp/util_gaussian.py
index e6696d3..fef0f15 100644
--- a/chesspp/util_gaussian.py
+++ b/chesspp/util_gaussian.py
@@ -1,12 +1,14 @@
 import math
+from typing import Tuple, Dict
+from functools import cache
 
 import torch
 import torch.distributions as dist
 from torch import exp
 
-F1: dict[float, float] = {}
-F2: dict[float, float] = {}
-CDF: dict[float, float] = {}
+F1: Dict[float, float] = {}
+F2: Dict[float, float] = {}
+CDF: Dict[float, float] = {}
 lookup_count = 0
 
 
@@ -15,7 +17,17 @@ def get_lookup_count():
     return lookup_count
 
 
-def max_gaussian(mu1, sigma1, mu2, sigma2) -> tuple[float, float]:
+@cache
+def calc_cdf(alpha: float) -> Tuple[float, float, float]:
+    normal = dist.Normal(0, 1)
+    cdf_alpha = normal.cdf(torch.tensor(alpha)).item()
+    pdf_alpha = exp(normal.log_prob(torch.tensor(alpha))).item()
+    f1 = alpha * cdf_alpha + pdf_alpha
+    f2 = alpha ** 2 * cdf_alpha * (1 - cdf_alpha) + (
+                    1 - 2 * cdf_alpha) * alpha * pdf_alpha - pdf_alpha ** 2
+    return cdf_alpha, f1, f2
+
+def max_gaussian(mu1, sigma1, mu2, sigma2) -> Tuple[float, float]:
     global lookup_count
     global F1
     global F2
@@ -31,33 +43,11 @@ def max_gaussian(mu1, sigma1, mu2, sigma2) -> tuple[float, float]:
     """
     # we assume independence of the two gaussians
     #print(mu1, sigma1, mu2, sigma2)
-    normal = dist.Normal(0, 1)
+    #normal = dist.Normal(0, 1)
     sigma_m = math.sqrt(sigma1 ** 2 + sigma2 ** 2)
-    alpha = (mu1 - mu2) / sigma_m
+    alpha = round((mu1 - mu2) / sigma_m, 2)
 
-    if alpha in CDF:
-        cdf_alpha = CDF[alpha]
-        lookup_count += 1
-    else:
-        cdf_alpha = normal.cdf(torch.tensor(alpha)).item()
-        CDF[alpha] = cdf_alpha
-
-    pdf_alpha = exp(normal.log_prob(torch.tensor(alpha))).item()
-
-    if alpha in F1:
-        f1_alpha = F1[alpha]
-        lookup_count += 1
-    else:
-        f1_alpha = alpha * cdf_alpha + pdf_alpha
-        F1[alpha] = f1_alpha
-
-    if alpha in F2:
-        f2_alpha = F2[alpha]
-        lookup_count += 1
-    else:
-        f2_alpha = alpha ** 2 * cdf_alpha * (1 - cdf_alpha) + (
-                    1 - 2 * cdf_alpha) * alpha * pdf_alpha - pdf_alpha ** 2
-        F2[alpha] = f2_alpha
+    cdf_alpha, f1_alpha, f2_alpha = calc_cdf(alpha)
 
     mu = mu2 + sigma_m * f1_alpha
     sigma = math.sqrt(sigma2 ** 2 + (sigma1 ** 2 - sigma2 ** 2) * cdf_alpha + sigma_m ** 2 * f2_alpha)
diff --git a/main.py b/main.py
index ef8d05c..a33262f 100644
--- a/main.py
+++ b/main.py
@@ -1,4 +1,5 @@
 import random
+import sys
 import time
 import chess
 import chess.engine
@@ -158,3 +159,8 @@ def main():
 
 if __name__ == '__main__':
     main()
+
+    # Note: prevent endless wait on StockFish process
+    # by allowing for cleanup of objects (which closes stockfish)
+    import gc
+    gc.collect()