-
Notifications
You must be signed in to change notification settings - Fork 0
/
joblb_comp_useprune.py
134 lines (109 loc) · 3.53 KB
/
joblb_comp_useprune.py
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
from analysis import *
import argparse
from matplotlib import rcParams
from pprint import pprint
rcParams.update(
{
"font.family": "serif",
"mathtext.fontset": "stix",
"font.serif": ["SimSun"],
"axes.unicode_minus": False,
}
)
ticklabel_style = {
"fontname": "Times New Roman",
}
parser = argparse.ArgumentParser(description="draw flow slowdown in each interval under various load")
parser.add_argument(
"-c",
"--csv",
dest="csv",
help="experiment configure name",
type=str,
metavar="csv_path",
required=True,
)
parser.add_argument(
"-l",
"--legend",
dest="legendname",
help="which legends to plot",
choices=["useprune", "K"],
type=str,
metavar="legend_name",
default="epsion",
)
args = parser.parse_args()
print("-" * 10, "reading data", "-" * 10)
sheet = read_csv(args.csv, use_cached=True)
_kept_rows = ["jobRCT:vector"]
jobs = get_vectors(
sheet,
names=_kept_rows,
module="FatTree",
)
runs = get_runIDs(sheet, by="iterationvars")
# pprint(runs)
assert isinstance(runs, dict)
# * a little hack to transpose the columns
jobs = jobs.set_index(["runID", "name"], drop=True)["vecvalue"].unstack() # pivot the name column
df = pd.DataFrame(columns=["load", "optprune", "jct"])
for itervar, repetition_ids in runs.items():
def _extract_iterationvar(iterationvar: str):
load = extract_float(iterationvar, "load")
useprune = extract_int(iterationvar, "optprune")
return load, useprune
load, useprune = _extract_iterationvar(itervar)
print(load, useprune)
jobreps = jobs.loc[repetition_ids]
jct = np.concatenate(jobreps["jobRCT:vector"].values)
df.loc[len(df.index)] = [load, useprune, jct] # pyright: ignore reportGeneralTypeIssues
jobs = df
loads = sorted(list(set([extract_float(x, "load") for x in runs.keys()])))
useprunes = sorted(list(set([extract_int(x, "optprune") for x in runs.keys()])))
fig, ax = plt.subplots(figsize=(12 / 2.54, 7 / 2.54))
_pos = np.arange(len(loads))
_bar_width = 0.2
bps = []
# if args.legendname == "useprune":
# legends = useprunes
# elif args.legendname == "K":
# legends =
compare_parm = useprunes
for step, param in enumerate(compare_parm):
current_load = jobs[jobs["optprune"] == param]
jct_mean = []
for load in loads:
x = current_load[current_load["load"] == load]
jct_mean.append(x["jct"].values[0].mean())
print(param, load, x["jct"].values[0].mean())
bp = ax.bar(_pos + step * _bar_width, jct_mean, _bar_width, hatch=HATCHES[step], color=COLORS[step], ec="black")
bps.append(bp)
plt.xticks(fontname="Times New Roman")
ax.tick_params(direction="in")
ax.set_xticks(_pos + _bar_width / 2, loads)
ax.legend(
[b[0] for b in bps],
["不使用改进策略","使用改进策略"],
# [f"useprune={legend}" for legend in compare_parm],
frameon=False,
prop={"family": "SimSun"},
ncol=2,
columnspacing=0.8,
loc="upper right",
bbox_to_anchor=(1, 1.04),
)
ax.set_ylim(ymin=0, ymax=0.035)
ax.set_xlabel("负载")
ax.set_ylabel("汇聚组平均每轮完成时间/s")
ax.set_axisbelow(True)
ax.yaxis.grid(color="gray", linestyle="dashed", alpha=0.5)
ax.xaxis.grid(color="gray", linestyle="dashed", alpha=0.5)
fig.subplots_adjust(left=0.12, bottom=0.15, right=0.99, top=0.98)
# fig.subplots_adjust(wspace=0.1)
output_name = "jobrct_useprune.png"
fig.savefig(output_name, dpi=600)
print(f"output {output_name}")
import os
paper_path = os.path.join(os.getcwd(), f"./{output_name}")
fig.savefig(paper_path, dpi=600)