1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
|
"""
Check the speed of 2d convolution.
"""
from __future__ import division, print_function, absolute_import
from itertools import product
import time
import numpy as np
from numpy.testing import Tester, TestCase, assert_allclose
from scipy.signal import convolve2d, correlate2d
def bench_convolve2d():
np.random.seed(1234)
# sample a bunch of pairs of 2d arrays
tm_start = time.clock()
pairs = []
for ma, na, mb, nb in product((1, 2, 8, 13, 30), repeat=4):
a = np.random.randn(ma, na)
b = np.random.randn(mb, nb)
pairs.append((a, b))
tm_end = time.clock()
tm_total = tm_end - tm_start
print('time to sample random pairs of 2d arrays:')
print(tm_total)
print()
fns = (convolve2d, correlate2d)
modes = ('full', 'valid', 'same')
boundaries = ('fill', 'wrap', 'symm')
# compute 2d convolutions and correlations for each 2d array pair
tm_start = time.clock()
for a, b in pairs:
for fn, mode, boundary in product(fns, modes, boundaries):
if mode == 'valid':
if b.shape[0] > a.shape[0] or b.shape[1] > a.shape[1]:
continue
fn(a, b, mode=mode, boundary=boundary)
tm_end = time.clock()
tm_total = tm_end - tm_start
print('time to compute 2d convolutions:')
print(tm_total)
print()
|
