-
Notifications
You must be signed in to change notification settings - Fork 12
/
parser_pdc_log.py
304 lines (270 loc) · 11.5 KB
/
parser_pdc_log.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
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
import matplotlib.pyplot as plt
import numpy as np
import sys
from os import walk
import matplotlib.colors as mcolors
'''
Return: table of lists of intervals {[%f,%f]}
'''
def read_log_file(filename):
result = {}
f = open(filename, 'r')
lines = f.readlines()
for line in lines:
line = line.replace('\n', '')
time_ranges = line.split(',')
entry_key = time_ranges[0]
result[entry_key] = []
for i in range(1, len(time_ranges)):
time_range = time_ranges[i].split("-")
result[entry_key].append([float(time_range[0]),float(time_range[1])])
'''
if len(result[entry_key]) == 0:
del result[entry_key]
'''
f.close()
return result
'''
Return: table of timings {%f}
'''
def read_timing_file(filename):
result = {}
f = open(filename, 'r')
lines = f.readlines()
for line in lines:
line = line.replace('\n', '')
timings = line.split(',')
entry_key = timings[0]
value = float(timings[1])
result[entry_key] = float(timings[1])
f.close()
return result
'''
Input: list of tables for lists of intervals [{[%f, %f]}]
Output: list of tables for timings [{%f}]
'''
def summarize_log_file(time_logs):
time_log_means = {}
time_log_counts = {}
time_log_max = {}
time_log_min = {}
keys = set()
for time_log in time_logs:
for k in time_log:
keys.add(k)
for time_log in time_logs:
for k in keys:
if k in time_log:
if k not in time_log_means:
time_log_counts[k] = 1
time_log_means[k] = time_log[k]
time_log_max[k] = time_log[k]
time_log_min[k] = time_log[k]
else:
time_log_counts[k] += 1
time_log_means[k] += time_log[k]
time_log_max[k] = max(time_log_max[k], time_log[k])
time_log_min[k] = min(time_log_min[k], time_log[k])
return [time_log_means, time_log_max, time_log_min, time_log_counts]
def interval_to_log(all_intervals):
results = []
for intervals in all_intervals:
next_timing = {}
for k in intervals:
next_timing[k] = .0
for i in range(0,len(intervals[k])):
next_timing[k] += intervals[k][i][1] - intervals[k][i][0]
results.append(next_timing)
return results
'''
Reshape the intervals
Input: Table for lists of intervals {[%f,%f]}
'''
def rescale_time(all_intervals, base):
for k in all_intervals:
for i in range(0, len(all_intervals[k])):
all_intervals[k][i] -= base
def check_interval_overlap(interval1, interval2):
return not (interval1[0] > interval2[1] or interval2[0] > interval1[1])
def check_all_intervals(input_intervals):
intervals = input_intervals[:]
intervals.sort(key=sort_by_lower_bound)
for i in range(1, len(intervals)):
if check_interval_overlap(intervals[i-1], intervals[i]):
return 1
return 0
'''
Intervals that overlap with each other will be merged.
'''
def coalesce_intervals(intervals):
intervals.sort(key=sort_by_lower_bound)
previous = 0
for i in range(1, len(intervals)):
if intervals[i][0] < intervals[previous][1]:
intervals[previous][1] = max(intervals[i][1], intervals[previous][1])
else:
previous += 1
intervals[previous][0] = intervals[i][0]
intervals[previous][1] = intervals[i][1]
if previous != len(intervals) - 1:
del intervals[-(len(intervals) - previous - 1):]
'''
Coalesce all intervals in each of the table entry.
Input: List of tables for interval lists [{[%f,%f]}]
Output: Same shape as intervals, but the inner list [%f,%f] are coalesced intervals
'''
def coalesce_all_intervals(all_intervals):
for intervals in all_intervals:
for k in intervals:
coalesce_intervals(intervals[k])
def sort_by_lower_bound(interval):
return interval[0]
def check_interval_gaps(input_intervals):
if input_intervals is None or len(input_intervals) ==0:
return None
intervals=input_intervals[:]
gaps = []
intervals.sort(key=sort_by_lower_bound)
previous = intervals[0][1]
for i in range(1, len(intervals)):
if previous < intervals[i][0]:
gaps.append((previous,intervals[i][0]))
previous = intervals[i][1]
return gaps
def total_interval_length(gaps):
return np.sum([(gap[1] - gap[0]) for gap in gaps])
def total_interval_std(gaps):
return np.std([(gap[1] - gap[0]) for gap in gaps])
def max_interval(gaps):
return np.max([(gap[1] - gap[0]) for gap in gaps])
def merge_intervals(all_intervals):
result = []
for k in all_intervals:
for interval in all_intervals[k]:
result.append(interval)
return result
def read_clients(n_clients, base, path):
result = []
for i in range(0, n_clients):
filename = '{0}/pdc_client_log_rank_{1}.csv'.format(path, i)
result.append(read_log_file(filename))
rescale_time(result[i], base)
return result
def plot_interval(all_intervals, index, key_name):
y = [index, index]
if key_name == 'transfer_request_start_write' or key_name == 'transfer_request_start_write_write':
color = 'red'
if key_name == 'transfer_request_wait_write' or key_name == 'transfer_request_wait_write_write':
color = 'yellow'
if key_name == 'transfer_request_wait_write_bulk':
color = 'green'
return 0
if key_name == 'transfer_request_inner_write_bulk':
color = 'black'
for interval in all_intervals:
plt.plot(interval, y, color = color)
def plot_all(server_intervals, client_intervals):
plt.figure()
plt.xlabel('time/sec')
for i in range(len(server_intervals)):
for interval in server_intervals[i]:
plot_interval(server_intervals[i][interval], 0, interval)
for i in range(0, len(client_intervals)):
for interval in client_intervals[i]:
plot_interval(client_intervals[i][interval], 1 + i, interval)
plt.savefig('{0}'.format("test_figure.pdf"))
plt.close()
def pdc_log_analysis(path):
print('====== Start analyzing path {0} ======'.format(path))
filenames = next(walk(path), (None, None, []))[2]
time_logs = []
interval_logs = []
for filename in filenames:
if 'pdc_server_timings' in filename:
full_filename = '{0}/{1}'.format(path, filename)
time_logs.append(read_timing_file(full_filename))
elif 'pdc_server_log_rank' in filename:
full_filename = '{0}/{1}'.format(path, filename)
interval_logs.append(read_log_file(full_filename))
time_log_means, time_log_max, time_log_min, time_log_counts = summarize_log_file(time_logs)
for k in time_log_means:
time_log_means[k] /= time_log_counts[k]
print('Key = {0}, mean = {1:.4}, min = {2:.4}, max = {3:.4}, count = {4}'.format(k, time_log_means[k], time_log_min[k], time_log_max[k], time_log_counts[k]))
base = np.min([np.min([server_intervals[x][0] for x in server_intervals if len(server_intervals[x]) > 0]) for server_intervals in interval_logs])
for server_intervals in interval_logs:
rescale_time(server_intervals, base)
coalesce_all_intervals(interval_logs)
interval_time_logs = interval_to_log(interval_logs)
time_log_means, time_log_max, time_log_min, time_log_counts = summarize_log_file(interval_time_logs)
print('start to print data from interval logs')
for k in time_log_means:
time_log_means[k] /= time_log_counts[k]
print('Key = {0}, mean = {1:.4}, min = {2:.4}, max = {3:.4}, count = {4}'.format(k, time_log_means[k], time_log_min[k], time_log_max[k], time_log_counts[k]))
return interval_time_logs, time_logs
def wrap_io_data(interval_time_logs, time_logs):
return np.mean([time_logs[i]['PDCreg_release_lock_bulk_transfer_inner_write_rpc'] + time_logs[i]['PDCregion_transfer_request_inner_write_bulk_rpc'] for i in range(0, len(time_logs))])
def wrap_comm_data(interval_time_logs, time_logs):
return np.mean([interval_time_logs[i]['transfer_request_wait_write_bulk'] + interval_time_logs[i]['release_lock_bulk_transfer_write'] for i in range(0, len(interval_time_logs))])
def wrap_other_data(interval_time_logs, time_logs):
return np.mean([time_logs[i]['PDCregion_transfer_start_write_rpc'] + time_logs[i]['PDCregion_transfer_wait_write_rpc'] + time_logs[i]['PDCreg_release_lock_write_rpc'] + time_logs[i]['PDCreg_obtain_lock_write_rpc'] for i in range(0, len(time_logs))] )
def compare_old_new_results(base_path, path):
path = "{0}shared_mode/vpic_old_results".format(base_path)
shared_old_interval_logs, shared_old_time_logs = pdc_log_analysis(path)
path = "{0}shared_mode/vpic_results".format(base_path)
shared_interval_logs, shared_time_logs = pdc_log_analysis(path)
path = "{0}dedicated_mode/vpic_old_results".format(base_path)
dedicated_old_interval_logs, dedicated_old_time_logs = pdc_log_analysis(path)
path = "{0}dedicated_mode/vpic_results".format(base_path)
dedicated_interval_logs, dedicated_time_logs = pdc_log_analysis(path)
all_interval_logs = [shared_old_interval_logs, shared_interval_logs, dedicated_old_interval_logs, dedicated_interval_logs]
all_time_logs = [shared_old_time_logs, shared_time_logs, dedicated_old_time_logs, dedicated_time_logs]
plt.figure()
width = 0.35
n_bars = 4
x_labels = np.arange(n_bars)
io_bar = [wrap_io_data(all_interval_logs[i], all_time_logs[i]) for i in range(0, n_bars)]
p_io_bar = plt.bar(x_labels, io_bar, width)
print("I/O bar")
print(io_bar)
comm_bar = [wrap_comm_data(all_interval_logs[i], all_time_logs[i]) for i in range(0, n_bars)]
p_comm_bar = plt.bar(x_labels, comm_bar, width, bottom=io_bar)
print("Comm bar")
print(comm_bar)
other_bar = [wrap_other_data(all_interval_logs[i], all_time_logs[i]) for i in range(0, n_bars)]
p_other_bar = plt.bar(x_labels, other_bar, width, bottom=[comm_bar[i]+io_bar[i] for i in range(0, n_bars)])
print("Other bar")
print(other_bar)
plt.ylabel('Timing/sec')
plt.title('Server Breakdown Timing Without Cache')
plt.legend((p_io_bar[0], p_comm_bar[0], p_other_bar[0]), ('I/O', 'Comm', 'Other'))
plt.xticks(x_labels, ('shared_bm', 'shared_tr', 'dedicated_bm', 'dedicated_tr'))
plt.savefig('{0}'.format("server_breakdown_no_cache.pdf"))
plt.close()
def analyze_single_log(path):
filenames = next(walk(path), (None, None, []))[2]
time_logs = []
interval_logs = []
for filename in filenames:
if 'pdc_server_timings' in filename:
full_filename = '{0}/{1}'.format(path, filename)
time_logs.append(read_timing_file(full_filename))
elif 'pdc_server_log_rank' in filename:
full_filename = '{0}/{1}'.format(path, filename)
interval_logs.append(read_log_file(full_filename))
print(len(interval_logs))
base = np.min([np.min([server_intervals[x][0] for x in server_intervals if len(server_intervals[x]) > 0]) for server_intervals in interval_logs])
for server_intervals in interval_logs:
rescale_time(server_intervals, base)
client_intervals = read_clients(31, base, path)
plot_all(interval_logs, client_intervals)
def main():
if len(sys.argv) == 2:
base_path = sys.argv[1]
if base_path[len(base_path) - 1] != '/':
base_path = '{0}/'.format(base_path)
else:
base_path = ''
#compare_old_new_results(base_path, path)
analyze_single_log(base_path)
if __name__== "__main__":
main()