Add memory perf benchmark

Author: urschrei

README.md

@@ -38,5 +38,13 @@ count = counter.calculate_final_result()
 # you can keep adding elements if you wish
 ```
 
+# Perf
+## Memory, 10e8 random 7-digit positive integers
+Allocation of integer array: 763 MiB
+
+`count_distinct`: 768 MiB
+
+`np.unique`: 1.7 GiB
+
 ## Note
 If you're thinking about using this library, you presumably know that it only provides an estimate (within the specified bounds), similar to something like HyperLogLog. You are trading accuracy for speed!

tests/bench_numpy.py

@@ -0,0 +1,16 @@
+import numpy as np
+import pyperf
+
+def add_elems(elems):
+    return np.unique(elems)
+
+
+# Generate 10 million random 7-digit integers
+random_integers = np.random.randint(low=1000000, high=10000000, size=10_000_000)
+
+runner = pyperf.Runner()
+runner.bench_func(
+    "10 million 7-digit random positive integers: numpy",
+    add_elems, random_integers
+)
+

tests/memory_count_distinct.py

@@ -0,0 +1,17 @@
+import numpy as np
+import pyperf
+from count_distinct import CVM
+
+
+def add_elems(counter, elems):
+    for elem in elems:
+        counter.add(elem)
+
+
+# Generate 10 million random 7-digit integers
+random_integers = np.random.randint(low=1000000, high=10000000, size=100_000_000)
+
+counter = CVM(0.8, 0.1, 100000)
+
+add_elems(counter, random_integers)
+count = counter.calculate_final_result()

tests/memory_numpy.py

@@ -0,0 +1,10 @@
+import numpy as np
+import memray
+
+def add_elems(elems):
+    return np.unique(elems)
+
+# Generate 100 million random 7-digit integers
+random_integers = np.random.randint(low=1000000, high=10000000, size=100_000_000)
+
+res = add_elems(random_integers)