def time_that(self, N, fun):
x = [0] * N
# Parallelized execution
start = time()
parallelize(x, fun, progressbar=False)
duration_multithreaded = time() - start
print(f"Multi-threaded execuition: {duration_multithreaded} seconds")
# Single-thread execution
start = time()
list([fun(i) for i in x])
duration_single = time() - start
print(f"Single-threaded execuition: {duration_single} seconds")
def test_series(self):
import pandas as pd
x = pd.Series([1, 2, 3], index=["a", "b", "c"])
y = parallelize(x, lambda p: p * 2)
y_true = x.apply(lambda p: p * 2)
self.assertTrue((y == y_true).all())
self.assertTrue((y.index == y_true.index).all())
def test_order(self):
def fun(x):
sleep(x)
return x
x = [0.5, 0.3, 0.2, 0.6, 0.4, 0.1, 0.5, 0.1, 0.2, 0.5, 0.6]
y = parallelize(x, fun)
self.assertEqual(x, y)
def test_exceptions(self):
x = [1, 2, 0, 5]
def fun(x):
if x == 0:
raise ZeroDivisionError
else:
return 10 / x
y = parallelize(x, fun)
self.assertEqual(y, [10, 5, None, 2])
try:
y = parallelize(x, fun, continue_on_exception=False)
self.assertTrue(False)
except ZeroDivisionError:
pass
obj = object()
y = parallelize(x, fun, exception_impute=obj)
self.assertEqual(y[2], obj)
def test_speed_increase(self):
x = [0.1, 0.1, 0.1, 0.1]
# Single-thread execution
start = time()
list([sleep(i) for i in x])
duration_single = time() - start
# Parallelized execution
start = time()
def fun(x):
sleep(x)
return x
y = parallelize(x, fun)
duration_multithreaded = time() - start
if multiprocessing.cpu_count() > 1:
# Strictly speaking, this might fail occasionally when
# some heavy process starts eating CPU right after single-threaded variant is executed.
self.assertLess(duration_multithreaded, duration_single)
def test_something(self):
x = [1, 2, 3, 4]
y = parallelize(x, lambda x: x * x)
self.assertEqual(y, [1, 4, 9, 16])
def test_empty(self):
e = parallelize([], lambda x: x)
self.assertEqual(len(e), 0)