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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324
|
#!/usr/bin/env python3
import sys
from pathlib import Path
from table import Table
class Input:
def __init__(self):
self.procedure = None
self.input_size = None
self.iterations = None
self.dataset = None
@property
def dataset_name(self):
if self.dataset:
return self.dataset.stem
else:
return 'none'
def __str__(self):
return '<Input: %s, size: %d, iterations: %s, dataset: %s>' % \
(self.procedure, self.input_size, self.iterations, self.dataset)
__repr__ = __str__
class Result:
def __init__(self):
self.instruction_per_byte = None
self.instruction_per_char = None
self.instruction_per_cycle = None
self.cycle_per_byte = None
self.cycle_per_char = None
self.bytes_per_char = None
self.speed_gigabytes = None
self.speed_gigachars = None
self.freq = None
self.time = None
def update_from_dict(self, D):
while D:
unit, value = D.popitem()
field = unit2field[unit]
if field is not None:
setattr(self, field, value)
unit2field = {
'ins/byte' : 'instruction_per_byte',
'ins/char' : 'instruction_per_char',
'ins/cycle' : 'instruction_per_cycle',
'cycle/byte' : 'cycle_per_byte',
'cycle/char' : 'cycle_per_char',
'byte/char' : 'byte_per_char',
'GHz' : 'freq',
'GB/s' : 'speed_gigabytes',
'Gc/s' : 'speed_gigachars',
'ns' : 'time',
'%' : None,
}
def parse(file):
result = []
for line in file:
for item in parse_line(line):
if isinstance(item, Input):
result.append([item, None])
else:
assert isinstance(item, dict)
if result[-1][1] is None:
result[-1][1] = Result()
result[-1][1].update_from_dict(item)
return [(input, result) for (input, result) in result if result is not None]
def parse_line(line):
if 'input size' in line:
yield parse_input(normalize_line(line))
elif 'GB/s' in line or 'ins/byte' in line or 'ins/char' in line:
yield parse_results(normalize_line(line))
def normalize_line(line):
line = line.replace(',', ' ')
line = line.replace(':', ' ')
line = line.replace('(', ' ')
line = line.replace(')', ' ')
return line.split()
def parse_input(fields):
input = Input()
input.procedure = fields.pop(0)
assert fields.pop(0) == 'input'
assert fields.pop(0) == 'size'
input.input_size = int(fields.pop(0))
assert fields.pop(0) == 'iterations'
input.iterations = int(fields.pop(0))
try:
assert fields.pop(0) == 'dataset'
input.dataset = Path(fields.pop(0))
except IndexError:
pass
return input
def try_float(s):
try:
return float(s)
except ValueError:
return
def parse_results(fields):
"""Consumes pairs "number unit" from the list"""
D = {}
while fields:
value = try_float(fields.pop(0))
if value is not None and fields:
D[fields.pop(0)] = value
return D
def speed_comparison(data):
procedures = set()
datasets = set()
results = {}
for input, result in data:
procedures.add(input.procedure)
datasets.add(input.dataset_name)
results[(input.procedure, input.dataset_name)] = result
datasets = list(sorted(datasets))
procedures = list(sorted(procedures))
def by_procedure():
table = Table()
table.set_header(['procedure'] + datasets)
for procedure in procedures:
row = []
row.append(procedure)
for dataset in datasets:
try:
result = results[(procedure, dataset)]
row.append('%0.3f GB/s' % (result.speed_gbs))
except KeyError:
row.append('--')
table.add_row(row)
return table
def by_dataset():
table = Table()
table.set_header(['dataset'] + procedures)
for dataset in datasets:
row = []
row.append(dataset)
for procedure in procedures:
try:
result = results[(procedure, dataset)]
row.append('%0.3f GB/s' % (result.speed_gigabytes))
except KeyError:
row.append('--')
table.add_row(row)
return table
if len(procedures) >= len(datasets):
return by_procedure()
else:
return by_dataset()
def compare(old, new):
collate = {}
for input, result in old:
key = str(input)
collate[key] = [input, result, None]
for input, result in new:
key = str(input)
if key not in collate:
collate[key] = [input, None, result]
else:
collate[key][-1] = result
grouped = {}
for _, (input, old, new) in collate.items():
if input.procedure not in grouped:
grouped[input.procedure] = []
grouped[input.procedure].append((input, old, new))
NA = '--'
table = Table()
table.set_header(["dataset", "size [B]", "iterations", "old GB/s", "new GB/s", "speedup"])
for procedure, values in grouped.items():
table.add_row([(procedure, 6)])
for input, old, new in values:
old_speed = old.speed_gigabytes
new_speed = new.speed_gigabytes
if old_speed is None or new_speed is None:
speedup = NA
else:
speedup = '%0.2fx' % (new_speed / old_speed)
if old_speed is None:
old_speed = NA
else:
old_speed = '%0.2f' % old_speed
if new_speed is None:
new_speed = NA
else:
new_speed = '%0.2f' % new_speed
table.add_row([
input.dataset.stem,
'%s' % input.input_size,
'%s' % input.iterations,
old_speed,
new_speed,
speedup,
])
return table
HELP = """Format output of the benchmark utility
When running a benchmark redirect its output to file (with > or `tee`), like:
$ cd build/benchmarks
$ ./benchmark [...] > results.txt
$ ./benchmark [...] | tee results.txt
To summarize results, use:
$ ./benchmark_print.py results.txt
| dataset | utf8_length_from_utf32+fallback | utf8_length_from_utf32+haswell | utf8_length_from_utf32+westmere |
| ---------------- | ------------------------------- | ------------------------------ | ------------------------------- |
| arabic.utf32 | 3.190 GB/s | 15.638 GB/s | 7.879 GB/s |
| chinese.utf32 | 3.179 GB/s | 15.349 GB/s | 7.644 GB/s |
| czech.utf32 | 3.178 GB/s | 15.721 GB/s | 7.801 GB/s |
...
| thai.utf32 | 3.191 GB/s | 15.455 GB/s | 7.842 GB/s |
| turkish.utf32 | 3.191 GB/s | 15.086 GB/s | 7.809 GB/s |
| vietnamese.utf32 | 3.182 GB/s | 15.317 GB/s | 7.807 GB/s |
To compare results, use:
$ ./benchmark_print.py old.txt new.txt
| dataset | size [B] | iterations | old GB/s | new GB/s | speedup |
| ------------------------------- | -------- | ---------- | -------- | -------- | ------- |
| convert_utf32_to_latin1+haswell |
| esperanto.utflatin32 | 328672 | 30000 | 77.14 | 88.81 | 1.15x |
| french.utflatin32 | 1729220 | 30000 | 75.26 | 92.39 | 1.23x |
| german.utflatin32 | 797324 | 30000 | 76.11 | 79.84 | 1.05x |
| portuguese.utflatin32 | 1086972 | 30000 | 84.12 | 92.64 | 1.10x |
"""
def parse_file(path):
with open(path, 'r') as f:
return parse(f)
def self_test():
rootdir = Path(__file__).parent / 'tests'
data = parse_file(rootdir / 'results.txt')
speed_comparison(data)
old = parse_file(rootdir / 'old.txt')
new = parse_file(rootdir / 'old.txt')
compare(new, old)
def main():
args = sys.argv[1:]
if "-h" in args or "--help" in args:
print(HELP)
return
if "--test" in args:
print("running tests...")
self_test()
print("all OK")
return
n = len(args)
if n == 1:
data = parse_file(args[0])
print(speed_comparison(data))
elif n == 2:
old = parse_file(args[0])
new = parse_file(args[1])
print(compare(old, new))
else:
print(HELP)
if __name__ == '__main__':
main()
|