ComputeLibrary/motivation.py at master · adityagupta1089/ComputeLibrary · 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
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import matplotlib
from collections import defaultdict

matplotlib.style.use("bmh")

data = {
    "AlexNet": """
CPU
2.04546, 0.204546, 10, 0
8.20984, 0.0820984, 10, 0
1.48004, 1.48004, 100, 0
GPU
1.68578, 0.168578, 0, 10
3.88249, 0.0388249, 0, 10
1.02315, 1.02315, 0, 100
CPU+GPU
1.08566, 0.108566, 5, 5
4.02829, 0.0402829, 3, 7
0.565674, 0.565674, 36, 64
""",
    "MobileNet": """
CPU
0.8054, 0.08054, 10, 0
7.26474, 0.0726474, 10, 0
0.165057, 0.165057, 100, 0
GPU
0.847036, 0.0847036, 0, 10
4.07821, 0.0407821, 0, 10
0.349245, 0.349245, 0, 100
CPU+GPU
0.597024, 0.0597024, 6, 4
3.06056, 0.0306056, 3, 7
0.121253, 0.121253, 70, 30
""",
    "GoogleNet": """
CPU
1.74059, 0.174059, 10, 0
16.394, 0.16394, 10, 0
0.40172, 0.40172, 100, 0
GPU
3.32846, 0.332846, 0, 10
8.56587, 0.0856587, 0, 10
0.908863, 0.908863, 0, 100
CPU+GPU
1.4559, 0.14559, 7, 3
5.32265, 0.0532265, 2, 8
0.292517, 0.292517, 70, 30
""",
    "SqueezeNet": """
CPU
0.903475, 0.0903475, 10, 0
8.74938, 0.0874938, 10, 0
0.154127, 0.154127, 100, 0
GPU
1.27485, 0.127485, 0, 10
4.3122, 0.043122, 0, 10
0.354992, 0.354992, 0, 100
CPU+GPU
0.876682, 0.0876682, 7, 3
3.23579, 0.0323579, 2, 8
0.12756, 0.12756, 79, 21
""",
    "ResNet50": """
CPU
4.28826, 0.428826, 10, 0
30.681, 0.30681, 10, 0
2.32899, 2.32899, 100, 0
GPU
3.94294, 0.394294, 0, 10
18.3556, 0.183556, 0, 10
1.94383, 1.94383, 100, 0
CPU+GPU
2.65938, 0.265938, 4, 6
12.9827, 0.129827, 2, 8
1.17527, 1.17527, 42, 58
""",
}

data2 = []
rows = [1, 2, 3, 5, 6, 7, 9, 10, 11]
rowvs = [
    (target, config)
    for target in ["CPU", "GPU", "CPU + GPU"]
    for config in [(10, 10), (10, 100), (1, 100)]
]
data3 = defaultdict(list)
for graph, val in data.items():
    print(graph)
    for (target, config), (idx, line) in zip(
        rowvs,
        [(a, b) for a, b in enumerate([x for x in val.split("\n") if x]) if a in rows],
    ):
        data3[config].append([graph, target, line.split(",")[1]])

for config in data3:
    df = pd.DataFrame(
        data3[config], columns=["Graph", "Target", "Time (sec/inference)",],
    )
    print(df)

    sns.catplot(
        data=df,
        x="Graph",
        y="Time (sec/inference)",
        hue="Target",
        legend=True,
        kind="bar",
        legend_out=False,
    )
    print(f"Generated fig {config}")
    plt.savefig(f"motivation_{config[0]}_{config[1]}.png")