def main(clock_type):
FORMAT = '%(asctime)-15s %(message)s'
logging.basicConfig(format=FORMAT, level=logging.DEBUG)
if _SCRPDIR == "":
logging.error("Could not find script directory")
return True
if clock_type.upper() not in _PARSE_CLOCK_MAP:
logging.error("Supplied clock type \"%s\" has no parsing method",
clock_type)
return True
logging.info("Compiling and adb pushing timing exec")
compile_and_push(clock_type.upper())
logging.info("Running with clock_type: %s", clock_type)
for cont_level in _CONTENTION_LEVELS:
logging.info("Running with contention_level: %s", cont_level)
try:
benchmark.Benchmark(clock_type, cont_level)
parse.Parse(_PARSE_CLOCK_MAP[clock_type.upper()],
join(_SCRPDIR, "logs", clock_type, cont_level))
except BenchmarkException as err:
logging.error("ERROR: BenchmarkException caught: %s", err)
return False
except ParseException as err:
logging.error("ERROR: ParseException caught: %s", err)
return False
return True
def testCreateTestsSingle(self):
b = benchmark.Benchmark([mocks.MockNameServer(mocks.GOOD_IP)],
test_count=1)
results = b.CreateTests(('A mail.google.com', ))
self.assertEquals(results, [['A', 'mail.google.com']])
# Oops, this isn't a real tuple.
self.assertRaises(AssertionError, b.CreateTests, 'google.com')
def _main(args: Args) -> int:
if args.no_warnings:
warnings.filterwarnings("ignore")
test = benchmark.Benchmark(bucket=args.bucket)
test.run()
return 0
def testCreateTestsChunkRecords(self):
b = benchmark.Benchmark([mocks.MockNameServer(mocks.GOOD_IP)],
test_count=100)
results = b.CreateTests(('A mail.google.com', 'CNAME test.live.com'),
select_mode='chunk')
self.assertEquals(
results, [['A', 'mail.google.com'], ['CNAME', 'test.live.com']])
def testCreateTestsWeighted(self):
b = benchmark.Benchmark([mocks.MockNameServer(mocks.GOOD_IP)],
test_count=1000)
results = b.CreateTests(('google.com', 'live.com'))
self.assertEquals(len(results), 1000)
self.assertTrue(('A', 'www.google.com.') in results)
caches = [x for x in results if 'cache' in x[1]]
self.assertTrue(len(caches) > 0 and len(caches) < 50)
def main():
image_path = input('>> Caminho para a imagem com extensão: ')
image = img_process.ImageProcess(image_path)
huffman_obj = huffman.Huffman(image.img_array)
save_data.SaveData('results/', image.img_array, huffman_obj.hist,
huffman_obj.codes)
bench = benchmark.Benchmark(image, image.img_array, huffman_obj.codes)
print('\nRESULTADOS PODEM SER VISUALIZADOS EM "/results"')
def testDigestion(self):
ns_list = (mocks.MockNameServer(mocks.GOOD_IP),
mocks.MockNameServer(mocks.PERFECT_IP),
mocks.MockNameServer(mocks.BROKEN_IP),
mocks.MockNameServer(mocks.SLOW_IP))
b = benchmark.Benchmark(ns_list)
good = ns_list[0].FakeAnswer(None)
bad = ns_list[0].FakeAnswer(None, no_answer=True)
b.results = {
ns_list[0]: [[('www.google.com.', 'A', 2.90, bad),
('google.com.', 'A', 9.80, good),
('www.google.com.', 'A', 9.90, good)],
[('www.google.com.', 'A', 9.90, bad),
('google.com.', 'A', 9.90, good),
('www.google.com.', 'A', 9.80, good)]],
ns_list[1]: [[('www.google.com.', 'A', 3.40, good),
('google.com.', 'A', 3.40, good),
('www.google.com.', 'A', 3.60, good)],
[('www.google.com.', 'A', 3.30, good),
('google.com.', 'A', 3.30, good),
('www.google.com.', 'A', 3.40, good)]],
ns_list[2]: [[('www.google.com.', 'A', 60, None),
('google.com.', 'A', 60, None),
('www.google.com.', 'A', 60, None)],
[('www.google.com.', 'A', 60, None),
('google.com.', 'A', 60, None),
('www.google.com.', 'A', 60, None)]],
ns_list[3]: [[('www.google.com.', 'A', 26.25, good),
('google.com.', 'A', 26.30, good),
('www.google.com.', 'A', 26.10, good)],
[('www.google.com.', 'A', 26.40, good),
('google.com.', 'A', 12.40, bad),
('www.google.com.', 'A', 26.80, good)]]
}
expected = []
averages = dict([(x[0].ip, x[1]) for x in b.ComputeAverages()])
self.assertEquals(averages[mocks.GOOD_IP], 8.7000000000000011)
self.assertEquals(averages[mocks.PERFECT_IP], 3.4000000000000004)
self.assertEquals(averages[mocks.BROKEN_IP], 60)
self.assertEquals(averages[mocks.SLOW_IP], 24.041666666666664)
expected = [('127.127.127.127', 3.2999999999999998),
('127.0.0.1', 9.80), ('9.9.9.9', 26.10),
('192.168.0.1', 60)]
fastest = [(x[0].ip, x[1]) for x in b.FastestNameServerResult()]
self.assertEquals(fastest, expected)
expected = [
(None, '####', 3.2999999999999998),
(None, '##########', 9.8000000000000007),
(None, '###########################', 26.100000000000001),
(None,
'############################################################',
60)
]
self.assertEquals(b._LowestLatencyAsciiChart(), expected)
def testFastest(self):
ns_list = (mocks.MockNameServer('X', name='X'),
mocks.MockNameServer('O', name='O'),
mocks.MockNameServer('U', name='U'))
b = benchmark.Benchmark(ns_list)
good = ns_list[0].FakeAnswer(None)
b.results = {
ns_list[0]:
[[('www.microsoft.com.', 'A', 3.0879974365234375, good),
('www.youku.com.', 'A', 2.2590160369873047, good),
('www.orkut.co.in.', 'A', 25.511980056762695, good),
('cache-9.ku6.com.', 'A', 1013.6392116546631, good),
('wsj.com.', 'A', 2.3639202117919922, good),
('www.imagevenue.com.', 'A', 2.6688575744628906, good),
('www.travian.ae.', 'A', 2.5160312652587891, good),
('www.fotolog.net.', 'A', 2.6750564575195312, good),
('www.torrentz.com.', 'A', 2.7811527252197266, good),
('www.wer-kennt-wen.de.', 'A', 2.7070045471191406, good)]],
ns_list[1]:
[[('www.microsoft.com.', 'A', 82.499980926513672, good),
('www.youku.com.', 'A', 81.991195678710938, good),
('www.orkut.co.in.', 'A', 82.377910614013672, good),
('cache-9.ku6.com.', 'A', 1141.1499977111816, good),
('wsj.com.', 'A', 84.334135055541992, good),
('www.imagevenue.com.', 'A', 84.282875061035156, good),
('www.travian.ae.', 'A', 84.036111831665039, good),
('www.fotolog.net.', 'A', 84.750175476074219, good),
('www.torrentz.com.', 'A', 84.517002105712891, good),
('www.wer-kennt-wen.de.', 'A', 83.980083465576172, good)]],
ns_list[2]:
[[('www.microsoft.com.', 'A', 12.507915496826172, good),
('www.youku.com.', 'A', 357.06806182861328, good),
('www.orkut.co.in.', 'A', 46.499967575073242, good),
('cache-9.ku6.com.', 'A', 697.60799407958984, good),
('wsj.com.', 'A', 87.159872055053711, good),
('www.imagevenue.com.', 'A', 11.99793815612793, good),
('www.travian.ae.', 'A', 11.492013931274414, good),
('www.fotolog.net.', 'A', 12.087106704711914, good),
('www.torrentz.com.', 'A', 12.598991394042969, good),
('www.wer-kennt-wen.de.', 'A', 11.770963668823242, good)]]
}
expected = [('G', 2.2590160369873047), ('U', 11.492013931274414),
('O', 81.991195678710938)]
fastest = [(x[0].ip, x[1]) for x in b.FastestNameServerResult()]
self.assertEquals(fastest, expected)
expected = [
('X', '##', 2.2590160369873047),
('U', '########', 11.492013931274414),
('O', '#####################################################',
81.991195678710938)
]
self.assertEquals(b._LowestLatencyAsciiChart(), expected)
def PrepareBenchmark(self):
"""Setup the benchmark object with the appropriate dataset."""
if len(self.nameservers) == 1:
thread_count = 1
else:
thread_count = self.options.benchmark_thread_count
self.bmark = benchmark.Benchmark(self.nameservers,
query_count=self.options.query_count,
run_count=self.options.run_count,
thread_count=thread_count,
status_callback=self.UpdateStatus)
def main():
# GPU utilization
CONFIG_PATH = "/project/datasets-API/benchmark-package/config/mnist.toml"
NUM_RUNS = 1
PATH_TO_CHROME_TRACES = "/project/datasets-API/benchmark-package/chrome-traces/test.json"
benchmark = bm.Benchmark('data_API', 'LowResFrameClassifier', CONFIG_PATH)
tot_acc, tot_time = 0, 0
for i in range(NUM_RUNS):
tot_time += benchmark.run('training', 500, profile=False)[0]
# tot_acc += benchmark.run('validation', 1000, profile=False)[1]
print("RESULTS: Train time=", tot_time / NUM_RUNS,
", Validation accuracy=", tot_acc / NUM_RUNS * 100)
def testRun(self):
ns_list = (mocks.MockNameServer(mocks.GOOD_IP),
mocks.MockNameServer(mocks.PERFECT_IP),
mocks.MockNameServer(mocks.BROKEN_IP),
mocks.MockNameServer(mocks.SLOW_IP))
b = benchmark.Benchmark(ns_list, test_count=3, run_count=2)
self.assertRaises(AssertionError, b.Run)
b.CreateTests(['A www.google.com'])
self.assertEquals(b.test_data,
[['A', 'www.google.com'], ['A', 'www.google.com'],
['A', 'www.google.com']])
b.Run()
ips_tested = sorted([x.ip for x in b.results])
expected = ['127.0.0.1', '127.127.127.127', '192.168.0.1', '9.9.9.9']
self.assertEquals(ips_tested, expected)
self.assertEquals(len(b.results[ns_list[0]]), 2)
self.assertEquals(len(b.results[ns_list[0]][0]), 3)
def main():
perfs = libraries.instantiate_libraries()
perfs_ab_sorted = sorted(perfs, key=lambda l: l.name)
bnch = benchmark.Benchmark(perfs_ab_sorted)
results = []
for title, config in JSONS.items():
with config["path"].open('rt', encoding='utf-8') as f:
cur_jsn = f.read()
cur_jsn = six.ensure_str(cur_jsn) # this is needed for python 2
result = bnch.run(cur_jsn, config.get("times"))
results.append((title, result))
chart_data = charts.from_results(
results, urls=[config['url'] for config in JSONS.values()])
with open('results.json', 'w') as f:
json.dump(chart_data, f)
def testNormalRun(self):
ns_list = (mocks.MockNameServer(mocks.GOOD_IP),
mocks.MockNameServer(mocks.PERFECT_IP),
mocks.MockNameServer(mocks.BROKEN_IP),
mocks.MockNameServer(mocks.SLOW_IP))
b = benchmark.Benchmark(ns_list, test_count=3, run_count=2)
b.CreateTests(['google.com', 'live.com'])
b.Run()
expected = ['127.0.0.1', '127.127.127.127', '192.168.0.1', '9.9.9.9']
averages = dict([(x[0].ip, x[1]) for x in b.ComputeAverages()])
self.assertEquals(len(averages), 4)
self.assertTrue(averages[mocks.GOOD_IP] >= 8)
self.assertTrue(averages[mocks.PERFECT_IP] <= 5)
self.assertTrue(averages[mocks.BROKEN_IP] >= 59)
self.assertTrue(averages[mocks.SLOW_IP] >= 20)
self.assertEquals(b.BestOverallNameServer(), ns_list[1])
self.assertEquals(b.NearestNameServers(count=2),
[ns_list[1], ns_list[0]])
#!/usr/bin/env python
from __future__ import absolute_import, division, print_function
# ----------------------------
# Imports for other modules --
# ----------------------------
from lsst.qserv.admin import commons
import benchmark
if __name__ == '__main__':
testdata_dir = '/datapool/tmp/loader_test/test'
out_dir = '/datapool/tmp/loader_tmp/test'
config = commons.read_user_config()
bench = benchmark.Benchmark(testdata_dir, out_dir)
bench.run()
def testEmptyRun(self):
ns_list = (mocks.MockNameServer(mocks.GOOD_IP), )
b = benchmark.Benchmark(ns_list, test_count=3, run_count=2)
self.assertRaises(AssertionError, b.Run)
os.path.join(cwd, '..', '..', '..', 'utils', 'benchmarks',
'tensorflow'))
import benchmark
logging.basicConfig(
level=logging.getLevelName("DEBUG"),
format='%(asctime)s %(name)s %(levelname)s %(message)s',
datefmt='%Y-%m-%d %H:%M:%S')
layer_dict = {}
utils = {"train": False}
module = benchmark.Benchmark(
partial(graph_builder, layer_dict),
inputs,
initializer,
add_args,
iteration_report,
partial(initializer_sess, layer_dict, utils),
)
# Make the sparse FC layer and insert into the layer dictionary
# for use by the builder functions:
options = benchmark.parse_opts(module, False)
np.random.seed(options.np_seed)
layer_dict['fc_gen'] = lambda: make_random_sparse_fc_layer(options)
if options.train:
utils["train"] = True
fc = layer_dict['fc_gen']()
print(
import benchmark game_list = benchmark.Benchmark().get_game() list(game_list)
firstDB = None
for group in groups:
if os.path.exists("chunkMap.txt"):
os.system("rm chunkMap.txt")
if len(group["DBs"]) == 0: continue
firstDB = group["DBs"][0]["name"]
firstTable = group["DBs"][0]["tables"][0]
first = True
for db in group["DBs"]:
data_dir = db["dir"]
dbName = db["name"]
temp_dir = os.path.join(data_dir, "temp")
config = commons.read_user_config()
bench = benchmark.Benchmark(data_dir, temp_dir, dbName)
bench.run()
# change css here
if not first:
cmd = "mysql -D qservCssData --port=%s --socket %s --user=%s --password=%s " % (
config['mysqld']['port'], config['mysqld']['socket'],
config['mysqld']['user'], config['mysqld']['pass'])
# Change all other tables
oldKey = "/DBS/%s/.packed.json" % dbName
newKey = "/DBS/%s/.packed.json" % firstDB
dbcmd = '-e "UPDATE qservCssData.kvData css1, qservCssData.kvData css2 SET css1.kvVal=css2.kvVal WHERE css1.kvKey=\'%s\' AND css2.kvVal=\'%s\';"' % (
oldKey, newKey)
print(cmd + dbcmd)
os.system(cmd + dbcmd)
first = False
import benchmark import numpy as np #Instancio la clase benchmark bench = benchmark.Benchmark() #Coges todas las funciones como una lista funciones = bench.getFunciones() #El tamaño del array puede ser el que quieras print(funciones[0](np.zeros(10))) print(funciones[5](np.zeros(30))) #Devuelve el número de funciones print(bench.getNumFunciones()) #Devuelve el límite inferior print(bench.getLimInf()) #Devuelve el límite superior print(bench.getLimSup()) #Devuelve el valor que da la funciones 2 y 7 en el óptimo print(bench.getFuncMinValor(2)) print(bench.getFuncMinValor(7)) #Devuelve un diccionario con el límite superior e inferior del dominio y con el valor que da la función 5 en el óptimo. print(bench.getInfo(5)) #Coge la función 3 y evalúa en [0,...,0] f3 = bench.getFuncion(3)
# val = random()
# line[index] = val
# break
# return line
if __name__ == "__main__":
trainpath, testpath, outputname, noise, bags, c, stop_criterion = main(
sys.argv[1:])
if not os.path.exists("results/"):
os.makedirs("results/")
with open("results/{}.csv".format(outputname), "w") as f:
f.write(
"c,stop criterion,bags,noise,classifier,accuracy,precision,recall,f1 score,learnT,testT \n"
)
f.close()
x_train, y_train, x_test, y_test = preprocessor(trainpath,
testpath,
noise,
labels=[0, 1])
ben = benchmark.Benchmark(x_train, y_train, x_test, y_test)
ben.run_benchmark(outputname, noise, bags, c, stop_criterion)
# to override the noise in parameter of the script and run the benchmark on different noise percentages
# noises = [0.1, 0.15, 0.2, 0.25, 0.5]
# cs = range(4, 12, 1)
# for k in cs:
# for n in noises:
# x_train, y_train, x_test, y_test = preprocessor2(trainpath, testpath, n, labels=[0, 1])
# ben = benchmark.Benchmark(x_train, y_train, x_test, y_test)
# ben.run_benchmark(outputname, n, bags, c, stop_criterion)
def benchmark(self, use_default=True, df_benchmark_stats_electricity=None,
df_benchmark_stats_fossil_fuel=None):
"""
This function add Benchmark instances for the current Building instance
:return:
"""
print("Start benchamrking")
if use_default:
df_sample_bench_stats_e = constants.Constants.df_sample_benchmark_stats_e
df_sample_bench_stats_f = constants.Constants.df_sample_benchmark_stats_f
else:
df_sample_bench_stats_e = df_benchmark_stats_electricity
df_sample_bench_stats_f = df_benchmark_stats_fossil_fuel
if (hasattr(self, "im_electricity") and hasattr(self.im_electricity, "coeffs")):
# Electricity
self.benchmark_HSL_e = benchmark.Benchmark('beta_hdd',
self.im_electricity.coeffs['hsl'],
df_bench_stats=df_sample_bench_stats_e,
valid=self.im_electricity.coeff_validation['hsl'])
self.benchmark_HCP_e = benchmark.Benchmark('beta_beth',
self.im_electricity.coeffs['hcp'],
df_bench_stats=df_sample_bench_stats_e,
valid=self.im_electricity.coeff_validation['hcp'])
self.benchmark_BASE_e = benchmark.Benchmark('beta_base',
self.im_electricity.coeffs['base'],
df_bench_stats=df_sample_bench_stats_e,
valid=self.im_electricity.coeff_validation['base'])
self.benchmark_CCP_e = benchmark.Benchmark('beta_betc',
self.im_electricity.coeffs['ccp'],
df_bench_stats=df_sample_bench_stats_e,
valid=self.im_electricity.coeff_validation['ccp'])
self.benchmark_CSL_e = benchmark.Benchmark('beta_cdd',
self.im_electricity.coeffs['csl'],
df_bench_stats=df_sample_bench_stats_e,
valid=self.im_electricity.coeff_validation['csl'])
self.benchmark_HSL_e.benchmark(plot=False)
self.benchmark_HCP_e.benchmark(plot=False)
self.benchmark_BASE_e.benchmark(plot=False)
self.benchmark_CCP_e.benchmark(plot=False)
self.benchmark_CSL_e.benchmark(plot=False)
else:
self.benchmark_HSL_e = None
self.benchmark_HCP_e = None
self.benchmark_BASE_e = None
self.benchmark_CCP_e = None
self.benchmark_CSL_e = None
if (hasattr(self, "im_fossil_fuel") and hasattr(self.im_fossil_fuel, "coeffs")):
# Need to add default fossil fuel in the constants module !!!
# Default benchmark stats will be used is no specific benchmark stats are provided
# Fossil fuel
self.benchmark_HSL_f = benchmark.Benchmark('beta_hdd',
self.im_fossil_fuel.coeffs['hsl'],
df_bench_stats=df_sample_bench_stats_f,
valid=self.im_fossil_fuel.coeff_validation['hsl'])
self.benchmark_HCP_f = benchmark.Benchmark('beta_beth',
self.im_fossil_fuel.coeffs['hcp'],
df_bench_stats=df_sample_bench_stats_f,
valid=self.im_fossil_fuel.coeff_validation['hcp'])
self.benchmark_BASE_f = benchmark.Benchmark('beta_base',
self.im_fossil_fuel.coeffs['base'],
df_bench_stats=df_sample_bench_stats_f,
valid=self.im_fossil_fuel.coeff_validation['base'])
self.benchmark_CCP_f = benchmark.Benchmark('beta_betc',
self.im_fossil_fuel.coeffs['ccp'],
df_bench_stats=df_sample_bench_stats_f,
valid=self.im_fossil_fuel.coeff_validation['ccp'])
self.benchmark_CSL_f = benchmark.Benchmark('beta_cdd',
self.im_fossil_fuel.coeffs['csl'],
df_bench_stats=df_sample_bench_stats_f,
valid=self.im_fossil_fuel.coeff_validation['csl'])
self.benchmark_HSL_f.benchmark(plot=False)
self.benchmark_HCP_f.benchmark(plot=False)
self.benchmark_BASE_f.benchmark(plot=False)
self.benchmark_CCP_f.benchmark(plot=False)
self.benchmark_CSL_f.benchmark(plot=False)
else:
self.benchmark_HSL_f = None
self.benchmark_HCP_f = None
self.benchmark_BASE_f = None
self.benchmark_CCP_f = None
self.benchmark_CSL_f = None
# Plot benchmark html sections
self.benchmarking_bar_hsl_e_html = benchmark.Benchmark.generate_benchmark_bar_html(self.benchmark_HSL_e)
self.benchmarking_bar_hcp_e_html = benchmark.Benchmark.generate_benchmark_bar_html(self.benchmark_HCP_e)
self.benchmarking_bar_base_e_html = benchmark.Benchmark.generate_benchmark_bar_html(self.benchmark_BASE_e)
self.benchmarking_bar_ccp_e_html = benchmark.Benchmark.generate_benchmark_bar_html(self.benchmark_CCP_e)
self.benchmarking_bar_csl_e_html = benchmark.Benchmark.generate_benchmark_bar_html(self.benchmark_CSL_e)
self.benchmarking_bar_hsl_f_html = benchmark.Benchmark.generate_benchmark_bar_html(self.benchmark_HSL_f)
self.benchmarking_bar_hcp_f_html = benchmark.Benchmark.generate_benchmark_bar_html(self.benchmark_HCP_f)
self.benchmarking_bar_base_f_html = benchmark.Benchmark.generate_benchmark_bar_html(self.benchmark_BASE_f)
self.benchmarking_bar_ccp_f_html = benchmark.Benchmark.generate_benchmark_bar_html(self.benchmark_CCP_f)
self.benchmarking_bar_csl_f_html = benchmark.Benchmark.generate_benchmark_bar_html(self.benchmark_CSL_f)
import six
import werkzeug
from flask import Flask, request, jsonify
from flask_cors import CORS
from werkzeug.utils import secure_filename
import benchmark
import libraries
from web import charts
app = Flask(__name__, static_folder=None)
app.config['MAX_CONTENT_LENGTH'] = 1024 * 1024 * 2
CORS(app) # todo: fix to receive only from https://jsonperf.com
BNCH = benchmark.Benchmark(
sorted(libraries.instantiate_libraries(), key=lambda l: l.name))
def allowed_file(filename):
return '.' in filename and \
filename.rsplit('.', 1)[1].lower() in ['json', 'txt']
@app.route("/python3", methods=["POST"])
@app.route("/python2", methods=["POST"])
def test():
user_file = request.files.get(
'file') # type: werkzeug.datastructures.FileStorage
if not user_file or user_file.filename == '':
return 'file is missing', 400
if t_type == "and_title":
and_id = "%s#%d" % (docid, i)
omim_title[and_id] = t
TITLES.write("%s\t%s\t#%s\t%s\n" %
(and_id, t_type, i, t))
matches = m.tag_text(and_id, t, t_type)
for te in matches:
MATCHES.write("%s\t%s\t%s\n" % (and_id, t, te))
i += 1
#m.tag_text(docid, text, "text")
TITLES.close()
MATCHES.close()
raw_mapping = m.get_mapping()
precision, recall, f1 = benchmark.Benchmark(
b, "omim_benchmark.tsv").get_performance(raw_mapping)
sys.stderr.write("Precision: %s Recall: %s F1: %s\n" %
(precision, recall, f1))
#sys.exit()
for docid in raw_mapping.iterkeys():
data = raw_mapping[docid]
if (data.score == None):
print "%s\t%s\t%s\t%s" % (docid, data.entity, "|".join(
data.synonyms), omim_title[docid])
else:
print "%f\t%s\t%s\t%s\t%s" % (data.score, docid, data.entity,
"|".join(data.synonyms),
omim_title[docid])
if __name__ == '__main__':
# Add benchmark module to path
cwd = os.path.dirname(os.path.abspath(inspect.stack()[0][1]))
os.sys.path.insert(
1,
os.path.join(cwd, '..', '..', '..', 'utils', 'benchmarks',
'tensorflow'))
import benchmark # noqa: E402
logging.basicConfig(
level=logging.getLevelName("DEBUG"),
format='%(asctime)s %(name)s %(levelname)s %(message)s',
datefmt='%Y-%m-%d %H:%M:%S')
module = benchmark.Benchmark(graph_builder, inputs,
tf.global_variables_initializer, add_args,
iteration_report)
opts = benchmark.parse_opts(module, False)
np.random.seed(opts.random_seed)
if opts.shards > 1:
raise NotImplementedError(
"--shards option has not been implemented with this example")
if opts.replicas > 1:
raise NotImplementedError(
"--replicas option has not been implemented with this example")
print(
f" Dynamic Sparse Transformer Feed-forward Layer {'Train' if opts.train else 'Inference'} Synthetic benchmark.\n"
f" Batch size {opts.batch_size}.\n"
import benchmark
import config_hnsw as hnsw
algo_field_names = ("M", "efConstruction")
algo_field_types = ("INTEGER", "INTEGER")
case_field_names = ("efSearch", )
case_field_types = ("INTEGER", )
bench = benchmark.Benchmark(algo_field_names, algo_field_types, \
case_field_names, case_field_types)
for M in hnsw.Ms:
for efConstruction in hnsw.efConstructions:
algo_fields = (M, efConstruction)
key = "HNSW%d-%d" % algo_fields
parameters = "efConstruction=%d" % efConstruction
case_fields = []
for efSearch in hnsw.efSearchs:
case_fields.append((efSearch, ))
bench.run(key, parameters, algo_fields, case_fields)
help="Layer hidden size")
parser.set_defaults(batches_per_step=1000, steps=5, shards=2)
return parser
def iteration_report(opts, time):
return "{:5f} items/sec".format(opts.batch_size * opts.batches_per_step / time)
if __name__ == '__main__':
sys.path.insert(1, '../../../utils/benchmarks/popart')
import benchmark
module = benchmark.Benchmark(
graph_builder,
add_args,
iteration_report
)
opts = benchmark.parse_opts(module)
opts.train = opts.mode == "train"
# Log Benchmark Message
print("Popart Multi-IPU {} Synthetic benchmark.\n"
" Batch size {}.\n"
" Batches per Step {}.\n"
" Steps {}.\n"
" {} IPUs."
.format(
{"infer": "Inference", "eval": "Evaluation", "train": "Training"}[opts.mode],
ArrayRenderer independent of display """
import time
import numpy as np
from punyty.vector import Vector3
from punyty.objects import Cube
from punyty.renderers import ArrayRenderer
from punyty.scene import Scene
import benchmark
if __name__ == '__main__':
width = 800
height = 800
target_array = np.zeros((height, width, 3))
scene = Scene()
cube = Cube()
scene.add_object(cube)
renderer = ArrayRenderer(target_array=target_array)
bench = benchmark.Benchmark(renderer)
while True:
t = time.time()
renderer.render(scene)
cube.rotate(Vector3(time.time(), 0, 0))
fps = bench.update(t)
if fps:
print(fps)
def iteration_report(opts, time):
return "{:5f} items/sec".format(opts.batch_size * opts.batches_per_step * opts.replicas / time)
if __name__ == '__main__':
# Add benchmark module to path
cwd = os.path.dirname(os.path.abspath(inspect.stack()[0][1]))
sys.path.insert(1, os.path.join(cwd, '..', '..', '..', 'utils',
'benchmarks', 'tensorflow'))
import benchmark
module = benchmark.Benchmark(
graph_builder,
inputs,
initializer,
add_args,
iteration_report
)
options = benchmark.parse_opts(module, False)
if options.shards > 1:
raise NotImplementedError(
"--shards option has not been implemented with this example")
# Log Benchmark Message
print("Multi-layer LSTM with a dense final layer, {} Benchmark.\n"
" Batch size {}.\n"
" Batches per Step {}.\n"
" Steps {}.\n"
from examples import search
import benchmark as bm
import cProfile as profile
from functools import wraps
def test_func():
for i in range(100):
x = i**3
if __name__ == '__main__':
benchmark = bm.Benchmark()
benchmark.set_funcs(search.search1, search.search2, search.search3)
benchmark.set_inputs(search.search_inputs, search.search_inputs2)
benchmark.compare(10)
code = compile(test_func, '<string>', 'exec')
profile.run(code)
