Initial commit

models/dist_statesaving/new_benchmark.py

@@ -0,0 +1,58 @@
+# New benchmark that only compares different methods in function of the state complexity
+# This benchmark works out of the context of a real simulation!
+
+# Complexity is defined as the number of attributes
+from copy import deepcopy
+import cPickle as pickle
+import time
+import random
+
+class AttributesState(object):
+    def __init__(self, complexity):
+        self.complexity = complexity
+        for f in xrange(complexity):
+            setattr(self, str(f), None)
+
+    def set_initial(self):
+        for f in xrange(complexity):
+            setattr(self, str(f), random.random())
+
+    def copy(self):
+        a = AttributesState(self.complexity)
+        for f in xrange(self.complexity):
+            setattr(a, str(f), getattr(self, str(f)))
+        return a
+
+class SizeState(object):
+    def __init__(self, complexity):
+        self.values = [None] * complexity
+        self.complexity = complexity
+
+    def set_initial(self):
+        self.values = [random.random() for _ in xrange(complexity)]
+
+    def copy(self):
+        a = SizeState(self.complexity)
+        a.values = list(self.values)
+        return a
+
+def benchmark(s, f, out):
+    samples = 1000
+    for c in range(0, 300, 5):
+        if s == "AttributesState":
+            state = AttributesState(c)
+        elif s == "SizeState":
+            state = SizeState(c)
+        start = time.time()
+        for _ in xrange(samples):
+            f(state)
+        t = (time.time() - start) / samples * 1000
+        print("%i %f" % (c, t))
+        out.write("%i %f\n" % (c, t))
+
+for s in ["AttributesState", "SizeState"]:
+    for f in [("deepcopy", lambda i: deepcopy(i)), ("pickle", lambda i: pickle.loads(pickle.dumps(i))), ("custom", lambda i: i.copy())]:
+        print("%s -- %s" % (s, f[0]))
+        out = open("%s_%s" % (s, f[0]), 'w')
+        benchmark(s, f[1], out)
+        out.close()