-
Notifications
You must be signed in to change notification settings - Fork 1
/
parse_mom_sis_results.py
144 lines (124 loc) · 4.84 KB
/
parse_mom_sis_results.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
135
136
137
138
139
140
141
142
143
# parse_mom_sis_results
from convert_time import seconds_per_day, seconds_per_hour
import numpy as np
import os
import pandas as pd
import sys
def parse_exp_seconds(log_file_name):
month_days = [
31, # Jan
28, # Feb
31, # Mar
30, # Apr
31, # May
30, # Jun
31, # Jul
31, # Aug
30, # Sep
31, # Oct
30, # Nov
31] # Dec
parsing_exp = True
parsing_months = True
parsing_days = True
parsing_hours = True
with open(log_file_name, "r") as log_file:
for text_line in log_file:
if parsing_exp and text_line.lstrip().startswith("&coupler_nml"):
while parsing_exp:
text_line = next(log_file)
if text_line.lstrip().startswith("months"):
parsing_months = False
months_split = text_line.split('=')
months = int(months_split[1].split(',')[0])
if text_line.lstrip().startswith("days"):
parsing_days = False
days_split = text_line.split('=')
days = float(days_split[1].split(',')[0])
if text_line.lstrip().startswith("hours"):
parsing_hours = False
hours_split = text_line.split('=')
hours = float(hours_split[1].split(',')[0])
parsing_exp = (
parsing_months or
parsing_days or
parsing_hours)
return float(
(sum(month_days[:months]) + days) * seconds_per_day +
(hours * seconds_per_hour))
def parse_seconds_per_time_step(log_file_name):
with open(log_file_name, "r") as log_file:
for text_line in log_file:
if text_line.lstrip().startswith("&ocean_model_nml"):
while True:
text_line = next(log_file)
if text_line.lstrip().startswith("dt_ocean"):
seconds_split = text_line.split('=')
return float(seconds_split[1].split(',')[0])
def parse_layout(log_file_name):
with open(log_file_name, "r") as log_file:
for text_line in log_file:
if text_line.lstrip().startswith("layout"):
layout_split = text_line.split('=')
layout = layout_split[1].split(',')
rows = int(layout[0])
cols = int(layout[1])
return rows, cols
def parse_npes(log_file_name):
with open(log_file_name, "r") as log_file:
for text_line in log_file:
if text_line.lstrip().startswith("MPP started"):
return int(text_line.split('=')[1])
def parse_times(log_file_name):
field_name_len = 36
parsing_times = False
times = dict()
with open(log_file_name, "r") as log_file:
for text_line in log_file:
if text_line.startswith("Total runtime"):
parsing_times = True
if parsing_times:
field_name = text_line[:field_name_len].strip()
field_value = float(text_line[field_name_len:])
times[field_name] = field_value
return times
def derive_speeds(times, exp_seconds):
speeds = dict()
for field_name in times:
time = times[field_name]
speed_field_name = field_name + " speed"
speeds[speed_field_name] = (
0.0
if time == 0.0 else
float(exp_seconds) / time)
return speeds
def parse_ocean_log_file(log_file_name):
result = dict()
log_basename_split = os.path.basename(log_file_name).split('.')
arch_prefix = log_basename_split[2]
result["arch"] = (
log_basename_split[3]
if arch_prefix == "noht" else
arch_prefix)
exp_seconds = parse_exp_seconds(log_file_name)
result["exp_seconds"] = exp_seconds
seconds_per_time_step = parse_seconds_per_time_step(log_file_name)
result["seconds_per_time_step"] = seconds_per_time_step
result["nbr_time_steps"] = int(exp_seconds / seconds_per_time_step)
rows, cols = parse_layout(log_file_name)
result["rows"] = rows
result["cols"] = cols
result["ncpus"] = parse_npes(log_file_name)
times = parse_times(log_file_name)
result.update(times)
speeds = derive_speeds(times, exp_seconds)
result.update(speeds)
result["Ocean speed per cpu"] = result["Ocean speed"] / result["ncpus"]
result["ocean_time_per_step"] = result["Ocean"] / result["nbr_time_steps"]
result["ocean_cpu_time_per_step"] = result["ocean_time_per_step"] * result["ncpus"]
return result
def parse_all(log_file_list):
log_list = list()
for log_file_name in log_file_list:
log_list.append(parse_ocean_log_file(log_file_name))
return pd.DataFrame(log_list)