tensorpy/benchmarks.py at master · Kaia15/tensorpy · GitHub

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
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
from tensor import Tensor
import numpy as np
import time

def benchmark_transpose():
    data = [[[x + y + z for z in range(200)] for y in range(200)] for x in range(200)]
    t = Tensor(data)
    np_array = np.array(data)
    axes = (1, 2, 0)

    start = time.perf_counter()
    t_transposed = t.transpose(axes)
    end = time.perf_counter()
    print(f"Tensor (sequential) transpose time: {end - start:.4f}s")

    start = time.perf_counter()
    t_mp_transposed = t.multi_processing_transpose(axes)
    end = time.perf_counter()
    print(f"Tensor (multi-processing) transpose time: {end - start:.4f}s")

    start = time.perf_counter()
    np_transposed = np.transpose(np_array, (1, 2, 0))
    end = time.perf_counter()
    print(f"Numpy flatten: {end - start:.4f}s")

def benchmark_c_flatten():
    print ("------------------------------------------------------------")
    data = [[[x + y + z for z in range(100)] for y in range(100)] for x in range(100)]
    t = Tensor(data)
    np_array = np.array(data)
    order = 'C'

    start = time.perf_counter()
    flat_seq = t.flatten(order)
    end = time.perf_counter()
    print(f"Tensor (sequential) C-flatten time: {end - start:.4f}s")

    start = time.perf_counter()
    flat_mp = t.multi_processing_flatten(order)
    end = time.perf_counter()
    print(f"Tensor (multi-processing) C-flatten time: {end - start:.4f}s")

    start = time.perf_counter()
    flat_np = np_array.flatten(order)
    end = time.perf_counter()
    print(f"Numpy C-flatten: {end - start:.4f}s")

def benchmark_f_flatten():
    print ("------------------------------------------------------------")
    data = [[[x + y + z for z in range(100)] for y in range(100)] for x in range(100)]
    t = Tensor(data)
    np_array = np.array(data)
    order = 'F'

    start = time.perf_counter()
    flat_seq = t.flatten(order)
    end = time.perf_counter()
    print(f"Tensor (sequential) F-flatten time: {end - start:.4f}s")

    start = time.perf_counter()
    flat_mp = t.multi_processing_flatten(order)
    end = time.perf_counter()
    print(f"Tensor (multi-processing) F-flatten time: {end - start:.4f}s")

    start = time.perf_counter()
    flat_np = np_array.flatten(order)
    end = time.perf_counter()
    print(f"Numpy F-flatten: {end - start:.4f}s")

def benchmark_arange():
    print ("------------------------------------------------------------")
    # For Tensor class
    # 1 arg: `end`
    start = time.perf_counter()
    t = Tensor.arange(100)
    end = time.perf_counter()
    print(f"Tensor arange(100) time: {end - start:.4f}s")

    # 2 args: `start` & `end`
    start = time.perf_counter()
    t = Tensor.arange(1,100,2)
    end = time.perf_counter()
    print(f"Tensor arange(1, 100, 2) time: {end - start:.4f}s")

    # 3 args: `start`, `end`, and `step`
    start = time.perf_counter()
    t = Tensor.arange(1.0, 100.0, 0.1)
    end = time.perf_counter()
    print(f"Tensor arange(1.0, 100.0, 0.1) time: {end - start:.4f}s")

    print ("||")

    # For Numpy
    # 1 arg: `end`
    start = time.perf_counter()
    t = np.arange(100)
    end = time.perf_counter()
    print(f"Numpy arange(100) time: {end - start:.4f}s")

    # 2 args: `start` & `end`
    start = time.perf_counter()
    t = np.arange(1, 100, 2)
    end = time.perf_counter()
    print(f"Numpy arange(1, 100, 2) time: {end - start:.4f}s")

    # 3 args: `start`, `end`, and `step`
    start = time.perf_counter()
    t = np.arange(1.0, 100.0, 0.1)
    end = time.perf_counter()
    print(f"Numpy arange(1, 100, 2) time: {end - start:.4f}s")

def benchmark_reshape():
    # simple test
    # For Tensor class
    print ("------------------------------------------------------------")
    a = [
            [
                [
                    [1, 2, 3],
                    [4, 5, 6]
                ],
                [
                    [7, 8, 9],
                    [10, 11, 12]
                ]
            ],
            [
                [
                    [13, 14, 15],
                    [16, 17, 18]
                ],
                [
                    [19, 20, 21],
                    [22, 23, 24]
                ]
            ]
        ]
    start = time.perf_counter()
    t1 = Tensor.reshape(a, (2, 3, 4), order='F')
    end = time.perf_counter()
    print (f"Tensor reshape time: {end - start:.4f}s")

    # For Numpy
    start = time.perf_counter()
    t1 = np.reshape(a, (2, 3, 4), order='F')
    end = time.perf_counter()
    print (f"Numpy reshape time: {end - start:.4f}s")

if __name__ == "__main__":
    benchmark_transpose()
    benchmark_c_flatten()
    benchmark_f_flatten()
    benchmark_arange()
    benchmark_reshape()