-
Notifications
You must be signed in to change notification settings - Fork 0
/
fsd.py
134 lines (117 loc) · 4.01 KB
/
fsd.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
import bisect
from analysis import *
import argparse
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="legends compared to plot",
choices=["epsion", "policy"],
type=str,
default="epsion",
)
parser.add_argument(
"-p",
"--percentile",
dest="percentile",
type=float,
default=0.95,
)
parser.add_argument(
"-t",
"--thresholdsmall",
dest="threshold",
type=float,
help="threshold(MB) to determine small flows",
default=10,
)
args = parser.parse_args()
output_name = "fsd_" + args.legendname + ".png"
percentile_lowerbound = args.percentile
threshold = args.threshold
print("-" * 10, "reading data", "-" * 10)
sheet = read_csv(args.csv, True)
_kept_rows = ["flowSize:vector", "fct:vector", "idealFct:vector", "jobRCT:vector"]
flows = get_vectors(
sheet,
names=_kept_rows,
module="FatTree",
)
runs = get_runIDs(sheet, by="iterationvars")
assert isinstance(runs, dict)
print("-" * 10, "align flow and job finish time", "-" * 10)
truncate_vectime(flows, runs)
print("-" * 10, "calc slowdown", "-" * 10)
df = get_flows_slowdown(flows, runs)
policies = sorted(list(set([extract_str(x, "aggPolicy") for x in runs.keys()])))
epsions = sorted(list(set([extract_float(x, "epsion") for x in runs.keys()])))
loads = sorted(list(set([extract_float(x, "load") for x in runs.keys()])))
legends = []
if args.legendname == "epsion":
legends = epsions
elif args.legendname == "policy":
legends = policies
plt.rcParams["font.family"] = "Serif"
fig, ax = plt.subplots(1, len(loads), figsize=(50 / 2.54, 10 / 2.54))
_pos = np.arange(2)
_bar_width = 0.2
for col_index, load in enumerate(loads):
current_load = df[(df["load"] == load)]
bps = []
for step, legend in enumerate(legends):
print(load, legend)
current_data = current_load[current_load[args.legendname] == legend]
flsz = current_data.iloc[0, :]["flowsize"]
flsz.sort()
x = [0, bisect.bisect_left(flsz, args.threshold * 1e6) , len(flsz)]
x95 = []
flsd: np.ndarray = current_data["slowdown"].values[0]
flsd_intv = []
flsd_intv_data = []
flct = []
for l, r in itertools.pairwise(x):
lb = round(l + (r - l) * percentile_lowerbound)
x95.append((lb, r))
if len(flsd[lb:r]) == 0:
print_error(f"inval too small: {lb},{r}")
exit()
data = flsd[lb:r]
flsd_intv.append(data.mean())
flsd_intv_data.append(data)
flct.append(len(data))
print(flct)
# bp = ax[col_index].plot(_pos, flsd_intv, color=COLORS[step], marker=MARKERS[step])
bp = ax[col_index].bar(_pos + step * _bar_width, flsd_intv, _bar_width, color=COLORS[step], ec="black")
bps.append(bp)
# bp = ax[col_index].boxplot(
# flsd_intv_data,
# False,
# "",
# widths=0.4,
# patch_artist=True,
# positions=_pos + step,
# boxprops=dict(facecolor=COLORS[step]),
# )
# * xticks set only once each ax
ax[col_index].set_xticks(_pos+_bar_width/2, [f"flowsize <= {args.threshold} MB", f"flowsize > {args.threshold} MB"])
# ax[col_index].legend([b["boxes"][0] for b in bps], epsions)
if args.legendname == "epsion":
ax[col_index].legend([b[0] for b in bps], [f"K={0 if legend == 0.0 else 10}" for legend in legends])
elif args.legendname == "policy":
ax[col_index].legend([b[0] for b in bps], [f"{args.legendname}={legend}" for legend in legends])
ax[col_index].set_xlabel(f"Load={load}")
ax[0].set_ylabel("FCT slow down")
fig.subplots_adjust(left=0.05, bottom=0.15, right=0.95, top=0.95)
fig.subplots_adjust(wspace=0.1)
fig.savefig(output_name)
print(f"output {output_name}")