def profile_cpu_check_connectivity_ratio(datasets=None,
queries=None,
methods=None,
feature=None,
input_dir='reuse'):
for dataset in datasets:
for method in methods:
print 'dataset=%s,method=%d, intersect_time/(intersect_time+condition_time)' % (
dataset, method)
for query in queries:
filename = input_dir + '/' + dataset + '_a' + \
str(method) + '_q' + str(query) + '_' + feature + '.txt'
t = extract_keywords_count(
filename, 'check_connectivity/check_constraints')
print t
print 'dataset=%s, intersect_time/elapsed_time' % dataset
for query in queries:
filename = input_dir + '/' + dataset + '_a' + \
str(method) + '_q' + str(query) + '_' + feature + '.txt'
connectivity_time = extract_keywords_time(
filename, 'intersect_time')
elapsed_time = extract_keywords_time(filename, 'elapsed_time')
ratio = float(connectivity_time) / float(elapsed_time)
# print connectivity_time + ',' + elapsed_time + ',' + str(ratio)
print ratio
def profile_gpu_reuse(datasets=None,
queries=None,
methods=None,
feature=None,
input_dir='reuse'):
time_keywords = [
'process_level_time', 'check_order_non_equality_time',
'check_connectivity_time', 'organize_batch_time',
'reuse_overhead_time', 'predicate_time', 'intersect_time',
'compact_time', 'iterate_gather_time', 'load_balance_search_time',
'memory_operations_time'
]
for dataset in datasets:
for method in methods:
print "dataset=%s,method=%d, in ms" % (dataset, method)
for query in queries:
filename = input_dir + '/' + dataset + '_a' + \
str(method) + '_q' + str(query) + '_' + feature + '.txt'
time_collection = []
for keyword in time_keywords:
t = extract_keywords_time(filename, keyword)
time_collection.append(t)
ss = ''
for i in range(len(time_collection)):
if i > 0:
ss += ','
ss += time_collection[i]
print ss
def profile_gpu_set_intersect_time_ratio(datasets=None,
queries=None,
methods=None,
feature=None,
input_dir='reuse'):
for dataset in datasets:
for method in methods:
print "dataset=%s,method=%d, in ms" % (dataset, method)
for query in queries:
filename = input_dir + '/' + dataset + '_a' + \
str(method) + '_q' + str(query) + '_' + feature + '.txt'
set_intersect_time = extract_keywords_time(
filename, "check_constraints_time")
total_time = extract_keywords_time(filename, "elapsed_time")
ratio = float(set_intersect_time) / float(total_time)
# print "%s,%s" % (set_intersect_time, total_time)
# print set_intersect_time
print total_time
def profile_gpu_memory_operation_overheads(datasets=None,
queries=None,
methods=None,
feature=None,
input_dir='reuse'):
for dataset in datasets:
for method in methods:
print 'dataset=%s,method=%d, memory_operations_time' % (dataset,
method)
for query in queries:
filename = input_dir + '/' + dataset + '_a' + \
str(method) + '_q' + str(query) + '_' + feature + '.txt'
t = extract_keywords_time(filename, 'memory_operations_time')
print t
print 'dataset=%s, elapsed_time' % dataset
for query in queries:
filename = input_dir + '/' + dataset + '_a' + \
str(method) + '_q' + str(query) + '_' + feature + '.txt'
t = extract_keywords_time(filename, 'elapsed_time')
print t
def exp_effect_reuse(datasets=None,
queries=None,
methods=None,
feature=None,
input_dir='reuse'):
for dataset in datasets:
for method in methods:
print 'dataset=%s,method=%d' % (dataset, method)
for query in queries:
filename = input_dir + '/' + dataset + \
'_a' + str(method) + "_q" + str(query)
if len(feature) > 0:
filename += '_' + feature
filename += '.txt'
t = extract_keywords_time(filename, 'elapsed_time')
num = float(t) / 1000.0
# return the time in second
print num
def profile_gpu_check_connectivity_ratio(datasets=None,
queries=None,
methods=None,
feature=None,
input_dir='reuse'):
for dataset in datasets:
for method in methods:
print 'dataset=%s,method=%d, intersect_time/(predicate_time+condition_time)' % (
dataset, method)
for query in queries:
filename = input_dir + '/' + dataset + '_a' + \
str(method) + '_q' + str(query) + '_' + feature + '.txt'
intersect_time = extract_keywords_time(filename,
'intersect_time')
predicate_time = extract_keywords_time(filename,
'predicate_time')
condition_time = extract_keywords_time(filename,
'condition_time')
ratio = float(intersect_time) / \
(float(predicate_time) + float(condition_time))
print ratio
print 'dataset=%s,predicate_time/(predicate_time+condition_time)' % dataset
for query in queries:
filename = input_dir + '/' + dataset + '_a' + \
str(method) + '_q' + str(query) + '_' + feature + '.txt'
predicate_time = extract_keywords_time(filename,
'predicate_time')
condition_time = extract_keywords_time(filename,
'condition_time')
ratio = float(predicate_time) / \
(float(predicate_time) + float(condition_time))
print ratio
print 'dataset=%s,intersect_time/elapsed_time' % dataset
for query in queries:
filename = input_dir + '/' + dataset + '_a' + \
str(method) + '_q' + str(query) + '_' + feature + '.txt'
intersect_time = extract_keywords_time(filename,
'intersect_time')
elapsed_time = extract_keywords_time(filename, 'elapsed_time')
ratio = float(intersect_time) / float(elapsed_time)
print ratio
def profile_phases(datasets=None,
queries=None,
methods=None,
feature=None,
keywords=None,
input_dir='reuse'):
print keywords
for dataset in datasets:
for method in methods:
print "dataset=%s,method=%d" % (dataset, method)
for query in queries:
filename = input_dir + '/' + dataset + '_a' + \
str(method) + "_q" + str(query) + "_" + feature + ".txt"
times = []
for keyword in keywords:
t = extract_keywords_time(filename, keyword)
times.append(t)
ss = ''
for i in range(len(times)):
if i > 0:
ss += ' '
ss += times[i]
print ss