def CmpNormalModelWithDataSample():
firsts, others, babies = Babies.PartitionBabies()
weights = Babies.GetWightList(babies)
pmf = Pmf.MakePmfFromList(weights)
mu = pmf.Mean()
var = pmf.Var(mu)
sigma = math.sqrt(var)
print("mu = {}, var = {}, sigma = {}".format(mu, var, sigma))
# 经验分布, 数据
cdf = Cdf.MakeCdfFromPmf(pmf, name='data')
myplot.cdf(cdf)
# u, sigma --> 误差函数计算 模型
xs, yy = pmf.Render()
ys = [erf.NormalCdf(x, mu, sigma) for x in xs]
myplot.Plot(xs, ys, label='Model')
myplot.Show()
myplot.Clf()
def save_plots(self, outdir, tag=None):
# CDFs
for cdf in AwazzaLogAnalyzer.CDFS:
myplot.cdf(cdf['data'](self), numbins=cdf['numbins'],\
xlabel=cdf['xlabel'],\
labels=cdf['labels'],\
filename=os.path.join(outdir, '%s_%s.pdf' % (tag, cdf['filename'])),\
**cdf['kwargs'])
# Data usage bar chart
bar_labels = ('From Origin', 'To Client', 'Cached', 'Compressed', 'Closed Early')
bar_values = (self._request_ints['kb-from-origin'] / 1000000.0,\
self._request_ints['kb-to-client'] / 1000000.0,\
self._request_ints['kb-cached'] / 1000000.0,\
self._request_ints['kb-compressed'] / 1000000.0,\
self._request_ints['kb-lost-client-closed'] / 1000000.0)
myplot.plot([bar_labels], [bar_values], type='bar', label_bars=True,\
ylabel='Data (GB)', bar_padding=0.5, barwidth=0.5,\
filename=os.path.join(outdir, '%s_%s.pdf' % (tag, 'data_usage')))
def main():
if len(args.hars) == 1 and os.path.isdir(args.hars[0]):
harpaths = glob.glob(args.hars[0] + '/*.har')
else:
harpaths = args.hars
sizes = []
num_zeros = 0
for harpath in harpaths:
# load HAR
try:
har = Har.from_file(harpath)
logging.info(har)
except HarError:
logging.exception('Error parsing HAR')
continue
# store object sizes in array
for obj in har.objects:
sizes.append(obj.size)
if obj.size == 0:
num_zeros += 1
print num_zeros, 'zeros'
print 'Mean:\t\t%f B' % numpy.mean(sizes)
print 'Min: ', numpy.min(sizes)
for percentile in (1, 10, 25, 50, 75, 90, 99):
print '%dth percentile:\t%f B' % (percentile,\
numpy.percentile(sizes, percentile))
print 'Max: ', numpy.max(sizes)
myplot.cdf([numpy.array(sizes)/1024.0], xlabel='Object Size (kB)',\
title='Object Sizes in Alexa Top 500', filename='./object_size_cdf.pdf')
myplot.cdf([sizes], xlabel='Object Size (kB)', xscale='log',\
title='Object Sizes in Alexa Top 500', filename='./object_size_cdf_log.pdf')
def main():
if len(args.hars) == 1 and os.path.isdir(args.hars[0]):
harpaths = glob.glob(args.hars[0] + '/*.har')
else:
harpaths = args.hars
sizes = []
num_zeros = 0
for harpath in harpaths:
# load HAR
try:
har = Har.from_file(harpath)
logging.info(har)
except HarError:
logging.exception('Error parsing HAR')
continue
# store object sizes in array
for obj in har.objects:
sizes.append(obj.size)
if obj.size == 0:
num_zeros += 1
print num_zeros, 'zeros'
print 'Mean:\t\t%f B' % numpy.mean(sizes)
print 'Min: ', numpy.min(sizes)
for percentile in (1, 10, 25, 50, 75, 90, 99):
print '%dth percentile:\t%f B' % (percentile,\
numpy.percentile(sizes, percentile))
print 'Max: ', numpy.max(sizes)
myplot.cdf([numpy.array(sizes)/1024.0], xlabel='Object Size (kB)',\
title='Object Sizes in Alexa Top 500', filename='./object_size_cdf.pdf')
myplot.cdf([sizes], xlabel='Object Size (kB)', xscale='log',\
title='Object Sizes in Alexa Top 500', filename='./object_size_cdf_log.pdf')
sys.path.append('../../tools/myplot')
import myplot
##
## context plots
##
mean_configs_per_user_across_contexts = []
with open('context_results', 'r') as f:
lines = f.readlines()
mean_configs_per_user_across_contexts = eval(lines[-2].strip())
f.closed
myplot.cdf([mean_configs_per_user_across_contexts ], height_scale=0.6,
xlabel='Mean # Unique Configurations per Context across 1000 Users',
filename='configs_per_user.pdf')
##
## conflict plots
##
mean_conflicts_per_app_across_contexts = []
num_contexts_per_app_with_any_conflict = []
with open('conflict_results2', 'r') as f:
lines = f.readlines()
mean_conflicts_per_app_across_contexts = eval(lines[-4].strip())
num_contexts_per_app_with_any_conflict = eval(lines[-2].strip())
f.closed
def plot_results(filename_to_results, filenames=None):
# use the filenames list to make sure we process files in order
# (so we can control the order of the series on the plot)
if not filenames: filenames = filename_to_results.keys()
filename_to_data = defaultdict(lambda: defaultdict(list))
fraction_data = []
fraction_labels = []
absolute_data = []
absolute_labels = []
mean_percents_by_size = []
mean_absolutes_by_size = []
mean_by_size_xs = []
mean_by_size_ys = []
mean_by_size_yerrs = []
mean_by_size_labels = []
for filename in filenames:
results = filename_to_results[filename]
for r in results:
if r.status == SUCCESS:
filename_to_data[filename]['both_success'].append(r.url)
filename_to_data[filename]['mean_percent_inflations'].append(r.https_mean / r.http_mean)
filename_to_data[filename]['mean_absolute_inflations'].append(r.https_mean - r.http_mean)
filename_to_data[filename]['median_percent_inflations'].append(r.https_median / r.http_median)
filename_to_data[filename]['median_absolute_inflations'].append(r.https_median - r.http_median)
if r.size:
filename_to_data[filename]['mean_percent_by_size'].append( (r.size/1000.0, r.https_mean / r.http_mean, r.http_stddev) )
filename_to_data[filename]['mean_absolute_by_size'].append( (r.size/1000.0, r.https_mean - r.http_mean, r.http_stddev) )
filename_to_data[filename]['mean_http_by_size'].append( (r.size/1000.0, r.http_mean, r.http_stddev) )
filename_to_data[filename]['mean_https_by_size'].append( (r.size/1000.0, r.https_mean, r.http_stddev) )
elif r.status == FAILURE_NO_HTTP:
filename_to_data[filename]['no_http'].append(r.url)
elif r.status == FAILURE_NO_HTTPS:
filename_to_data[filename]['no_https'].append(r.url)
else:
filename_to_data[filename]['other_error'].append(r.url)
print '%i sites were accessible over both protocols' %\
len(filename_to_data[filename]['both_success'])
print '%i sites were not accessible over HTTP' %\
len(filename_to_data[filename]['no_http'])
print '%i sites were not accessible over HTTPS' %\
len(filename_to_data[filename]['no_https'])
print '%i sites were not accessible for other reasons' %\
len(filename_to_data[filename]['other_error'])
if 'pit' in filename:
location = 'PIT'
elif '3g' in filename:
location = '3G'
else:
location = 'Fiber'
fraction_data.append(filename_to_data[filename]['mean_percent_inflations'])
fraction_labels.append('Mean (%s)' % location)
fraction_data.append(filename_to_data[filename]['median_percent_inflations'])
fraction_labels.append('Median (%s)' % location)
absolute_data.append(numpy.array(filename_to_data[filename]['mean_absolute_inflations']))# * 1000) # s -> ms
absolute_labels.append('Mean (%s)' % location)
absolute_data.append(numpy.array(filename_to_data[filename]['median_absolute_inflations']))# * 1000) # s -> ms
absolute_labels.append('Median (%s)' % location)
try:
mean_by_size_xs.append(zip(*sorted(filename_to_data[filename]['mean_http_by_size']))[0])
mean_by_size_ys.append(zip(*sorted(filename_to_data[filename]['mean_http_by_size']))[1])
mean_by_size_yerrs.append(zip(*sorted(filename_to_data[filename]['mean_http_by_size']))[2])
mean_by_size_labels.append('Mean HTTP (%s)' % location)
mean_by_size_xs.append(zip(*sorted(filename_to_data[filename]['mean_https_by_size']))[0])
mean_by_size_ys.append(zip(*sorted(filename_to_data[filename]['mean_https_by_size']))[1])
mean_by_size_yerrs.append(zip(*sorted(filename_to_data[filename]['mean_https_by_size']))[2])
mean_by_size_labels.append('Mean HTTPS (%s)' % location)
except Exception as e:
logging.warn('Error processing size data: %s' % e)
if location == 'BCN':
mean_percents_by_size.append(filename_to_data[filename]['mean_percent_by_size'])
mean_absolutes_by_size.append(filename_to_data[filename]['mean_absolute_by_size'])
myplot.cdf(fraction_data,
xlabel='Load Time Ratio (HTTPS/HTTP)', labels=fraction_labels,
filename=os.path.join(args.outdir, '%s_fraction_inflation.pdf' % args.tag),
height_scale=0.7, numbins=10000, xlim=(1, 3), legend='lower right')
myplot.cdf(absolute_data,
xlabel='Load Time Difference (HTTPS-HTTP) [s]', labels=absolute_labels,
filename=os.path.join(args.outdir, '%s_absolute_inflation.pdf' % args.tag),
height_scale=0.7, numbins=10000, xlim=(0,3), legend='lower right')
myplot.cdf(absolute_data,
xlabel='Load Time Difference (HTTPS-HTTP) [s]', labels=absolute_labels,
filename=os.path.join(args.outdir, '%s_absolute_inflation_log.pdf' % args.tag),
height_scale=0.7, numbins=10000, xscale='log', xlim=(0, 10), legend='lower right')
# Plot fraction and absolute in same figure as subplots
fig, ax_array = myplot.subplots(1, 2, height_scale=0.75, width_scale=1.2)
myplot.cdf(fraction_data, fig=fig, ax=ax_array[0],
xlabel='Load Time Ratio\n(HTTPS/HTTP)', labels=fraction_labels,
numbins=10000, xlim=(1, 3), show_legend=False)
lines, labels = myplot.cdf(absolute_data, fig=fig, ax=ax_array[1],
xlabel='Load Time Difference\n(HTTPS-HTTP) [s]', labels=absolute_labels,
numbins=10000, xlim=(0,3), legend='lower right', labelspacing=0.1, handletextpad=0.4)
# shrink plots to make room for legend underneath
#for ax in ax_array:
# box = ax.get_position()
# ax.set_position([box.x0, box.y0 + box.height * 0.25,
# box.width, box.height * 0.75])
# shrink plots to make room for title above
for ax in ax_array:
box = ax.get_position()
ax.set_position([box.x0, box.y0,
box.width, box.height * 0.95])
#myplot.save_plot(os.path.join(args.outdir, '%s_combined_inflation_no_legend.pdf' % args.tag))
#fig.legend(lines, labels, loc='lower center', ncol=2, prop={'size':20}, frameon=False,
# bbox_to_anchor=(.5, -.03))
fig.suptitle('O-Proxy Top 2000 Objects')
myplot.save_plot(os.path.join(args.outdir, '%s_combined_inflation.pdf' % args.tag))
try:
myplot.plot([zip(*mean_percents_by_size[0])[0]], [zip(*mean_percents_by_size[0])[1]],
xlabel='Object Size (KB)', ylabel='Fraction Inflation (HTTPS/HTTP)',
linestyles=[''], xscale='log',
filename=os.path.join(args.outdir, '%s_fraction_by_size.pdf' % args.tag))
myplot.plot([zip(*mean_absolutes_by_size[0])[0]], [zip(*mean_absolutes_by_size[0])[1]],
xlabel='Object Size (KB)', ylabel='Absolute Inflation (HTTPS-HTTP) [sec]',
linestyles=[''], xscale='log',
filename=os.path.join(args.outdir, '%s_absolute_by_size.pdf' % args.tag))
myplot.plot(mean_by_size_xs, mean_by_size_ys, yerrs=mean_by_size_yerrs,
xlabel='Object Size (KB)', ylabel='Load Time [sec]', xscale='log',
marker=None, labels=mean_by_size_labels,# legend='lower left',
legend_cols=2, width_scale=2,
filename=os.path.join(args.outdir, '%s_mean_lt_by_size.pdf' % args.tag))
except Exception as e:
logging.warn('Error processing size data: %s', e)
def plot_results(filename_to_results, filenames=None):
# use the filenames list to make sure we process files in order
# (so we can control the order of the series on the plot)
if not filenames: filenames = filename_to_results.keys()
filename_to_data = defaultdict(lambda: defaultdict(list))
fraction_data = []
fraction_labels = []
absolute_data = []
absolute_labels = []
mean_percents_by_size = []
mean_absolutes_by_size = []
mean_by_size_xs = []
mean_by_size_ys = []
mean_by_size_yerrs = []
mean_by_size_labels = []
for filename in filenames:
results = filename_to_results[filename]
for r in results:
if r.status == SUCCESS:
filename_to_data[filename]['both_success'].append(r.url)
filename_to_data[filename]['mean_percent_inflations'].append(
r.https_mean / r.http_mean)
filename_to_data[filename]['mean_absolute_inflations'].append(
r.https_mean - r.http_mean)
filename_to_data[filename]['median_percent_inflations'].append(
r.https_median / r.http_median)
filename_to_data[filename][
'median_absolute_inflations'].append(r.https_median -
r.http_median)
if r.size:
filename_to_data[filename]['mean_percent_by_size'].append(
(r.size / 1000.0, r.https_mean / r.http_mean,
r.http_stddev))
filename_to_data[filename]['mean_absolute_by_size'].append(
(r.size / 1000.0, r.https_mean - r.http_mean,
r.http_stddev))
filename_to_data[filename]['mean_http_by_size'].append(
(r.size / 1000.0, r.http_mean, r.http_stddev))
filename_to_data[filename]['mean_https_by_size'].append(
(r.size / 1000.0, r.https_mean, r.http_stddev))
elif r.status == FAILURE_NO_HTTP:
filename_to_data[filename]['no_http'].append(r.url)
elif r.status == FAILURE_NO_HTTPS:
filename_to_data[filename]['no_https'].append(r.url)
else:
filename_to_data[filename]['other_error'].append(r.url)
print '%i sites were accessible over both protocols' %\
len(filename_to_data[filename]['both_success'])
print '%i sites were not accessible over HTTP' %\
len(filename_to_data[filename]['no_http'])
print '%i sites were not accessible over HTTPS' %\
len(filename_to_data[filename]['no_https'])
print '%i sites were not accessible for other reasons' %\
len(filename_to_data[filename]['other_error'])
if 'pit' in filename:
location = 'PIT'
elif '3g' in filename:
location = '3G'
else:
location = 'Fiber'
fraction_data.append(
filename_to_data[filename]['mean_percent_inflations'])
fraction_labels.append('Mean (%s)' % location)
fraction_data.append(
filename_to_data[filename]['median_percent_inflations'])
fraction_labels.append('Median (%s)' % location)
absolute_data.append(
numpy.array(filename_to_data[filename]
['mean_absolute_inflations'])) # * 1000) # s -> ms
absolute_labels.append('Mean (%s)' % location)
absolute_data.append(
numpy.array(filename_to_data[filename]
['median_absolute_inflations'])) # * 1000) # s -> ms
absolute_labels.append('Median (%s)' % location)
try:
mean_by_size_xs.append(
zip(*sorted(filename_to_data[filename]['mean_http_by_size']))
[0])
mean_by_size_ys.append(
zip(*sorted(filename_to_data[filename]['mean_http_by_size']))
[1])
mean_by_size_yerrs.append(
zip(*sorted(filename_to_data[filename]['mean_http_by_size']))
[2])
mean_by_size_labels.append('Mean HTTP (%s)' % location)
mean_by_size_xs.append(
zip(*sorted(filename_to_data[filename]['mean_https_by_size']))
[0])
mean_by_size_ys.append(
zip(*sorted(filename_to_data[filename]['mean_https_by_size']))
[1])
mean_by_size_yerrs.append(
zip(*sorted(filename_to_data[filename]['mean_https_by_size']))
[2])
mean_by_size_labels.append('Mean HTTPS (%s)' % location)
except Exception as e:
logging.warn('Error processing size data: %s' % e)
if location == 'BCN':
mean_percents_by_size.append(
filename_to_data[filename]['mean_percent_by_size'])
mean_absolutes_by_size.append(
filename_to_data[filename]['mean_absolute_by_size'])
myplot.cdf(fraction_data,
xlabel='Load Time Ratio (HTTPS/HTTP)',
labels=fraction_labels,
filename=os.path.join(args.outdir,
'%s_fraction_inflation.pdf' % args.tag),
height_scale=0.7,
numbins=10000,
xlim=(1, 3),
legend='lower right')
myplot.cdf(absolute_data,
xlabel='Load Time Difference (HTTPS-HTTP) [s]',
labels=absolute_labels,
filename=os.path.join(args.outdir,
'%s_absolute_inflation.pdf' % args.tag),
height_scale=0.7,
numbins=10000,
xlim=(0, 3),
legend='lower right')
myplot.cdf(absolute_data,
xlabel='Load Time Difference (HTTPS-HTTP) [s]',
labels=absolute_labels,
filename=os.path.join(
args.outdir, '%s_absolute_inflation_log.pdf' % args.tag),
height_scale=0.7,
numbins=10000,
xscale='log',
xlim=(0, 10),
legend='lower right')
# Plot fraction and absolute in same figure as subplots
fig, ax_array = myplot.subplots(1, 2, height_scale=0.75, width_scale=1.2)
myplot.cdf(fraction_data,
fig=fig,
ax=ax_array[0],
xlabel='Load Time Ratio\n(HTTPS/HTTP)',
labels=fraction_labels,
numbins=10000,
xlim=(1, 3),
show_legend=False)
lines, labels = myplot.cdf(absolute_data,
fig=fig,
ax=ax_array[1],
xlabel='Load Time Difference\n(HTTPS-HTTP) [s]',
labels=absolute_labels,
numbins=10000,
xlim=(0, 3),
legend='lower right',
labelspacing=0.1,
handletextpad=0.4)
# shrink plots to make room for legend underneath
#for ax in ax_array:
# box = ax.get_position()
# ax.set_position([box.x0, box.y0 + box.height * 0.25,
# box.width, box.height * 0.75])
# shrink plots to make room for title above
for ax in ax_array:
box = ax.get_position()
ax.set_position([box.x0, box.y0, box.width, box.height * 0.95])
#myplot.save_plot(os.path.join(args.outdir, '%s_combined_inflation_no_legend.pdf' % args.tag))
#fig.legend(lines, labels, loc='lower center', ncol=2, prop={'size':20}, frameon=False,
# bbox_to_anchor=(.5, -.03))
fig.suptitle('O-Proxy Top 2000 Objects')
myplot.save_plot(
os.path.join(args.outdir, '%s_combined_inflation.pdf' % args.tag))
try:
myplot.plot([zip(*mean_percents_by_size[0])[0]],
[zip(*mean_percents_by_size[0])[1]],
xlabel='Object Size (KB)',
ylabel='Fraction Inflation (HTTPS/HTTP)',
linestyles=[''],
xscale='log',
filename=os.path.join(
args.outdir, '%s_fraction_by_size.pdf' % args.tag))
myplot.plot([zip(*mean_absolutes_by_size[0])[0]],
[zip(*mean_absolutes_by_size[0])[1]],
xlabel='Object Size (KB)',
ylabel='Absolute Inflation (HTTPS-HTTP) [sec]',
linestyles=[''],
xscale='log',
filename=os.path.join(
args.outdir, '%s_absolute_by_size.pdf' % args.tag))
myplot.plot(
mean_by_size_xs,
mean_by_size_ys,
yerrs=mean_by_size_yerrs,
xlabel='Object Size (KB)',
ylabel='Load Time [sec]',
xscale='log',
marker=None,
labels=mean_by_size_labels, # legend='lower left',
legend_cols=2,
width_scale=2,
filename=os.path.join(args.outdir,
'%s_mean_lt_by_size.pdf' % args.tag))
except Exception as e:
logging.warn('Error processing size data: %s', e)
def compare_results(result_files):
'''Takes pickled result files (produced by analyze_traces) and plots stuff'''
filename_to_plt_list_dict = {} # filename -> url -> list of PLTs (seconds)
filename_to_size_list_dict = {} # filename -> url -> list of sizes (bytes)
for result_file in result_files:
with open(result_file, 'r') as f:
url_to_plts, url_to_sizes = cPickle.load(f)
f.closed
filename_to_plt_list_dict[result_file] = url_to_plts
filename_to_size_list_dict[result_file] = url_to_sizes
# collapse stats for each URL into mean and median
filename_to_plt_mean_dict = defaultdict(dict) #filename -> url -> mean PLT
filename_to_plt_median_dict = defaultdict(dict) #filename -> url -> median PLT
for filename, plt_list_dict in filename_to_plt_list_dict.iteritems():
for url, plt_list in plt_list_dict.iteritems():
filename_to_plt_mean_dict[filename][url] = numpy.mean(plt_list)
filename_to_plt_median_dict[filename][url] = numpy.median(plt_list)
filename_to_size_mean_dict = defaultdict(dict) # filename -> url -> mean size
filename_to_size_median_dict = defaultdict(dict) # filename -> url -> median size
for filename, size_list_dict in filename_to_size_list_dict.iteritems():
for url, size_list in size_list_dict.iteritems():
filename_to_size_mean_dict[filename][url] = numpy.mean(size_list) / 1000000.0 # bytes -> MB
filename_to_size_median_dict[filename][url] = numpy.median(size_list) / 1000000.0 # bytes -> MB
# plot CDFs
mean_plts = []
median_plts = []
mean_sizes = []
median_sizes = []
labels = []
for filename in result_files:
name = os.path.splitext(os.path.split(filename)[1])[0]
labels.append(string.replace(string.replace(name, 'SPDY', 'Compression Proxy'), 'NoProxy', 'No Proxy'))
mean_plts.append(filename_to_plt_mean_dict[filename].values())
median_plts.append(filename_to_plt_median_dict[filename].values())
mean_sizes.append(filename_to_size_mean_dict[filename].values())
median_sizes.append(filename_to_size_median_dict[filename].values())
# mean PLTs
myplot.cdf(mean_plts, height_scale=0.7,
xlabel='Mean Page Load Time (seconds)', labels=labels,
filename=os.path.join(args.outdir, 'mean_plt.pdf'))
myplot.cdf(median_plts, height_scale=0.7,
xlabel='Median Page Load Time (seconds)', labels=labels,
filename=os.path.join(args.outdir, 'median_plt.pdf'))
myplot.cdf(mean_sizes, height_scale=0.7,
xlabel='Mean Total Data Exchanged (MB)', labels=labels,
xlim=(0, 5),
filename=os.path.join(args.outdir, 'mean_size.pdf'))
myplot.cdf(median_sizes, height_scale=0.7,
xlabel='Median Total Data Exchanged (MB)', labels=labels,
xlim=(0, 5),
filename=os.path.join(args.outdir, 'median_size.pdf'))
combined_sizes = []
combined_labels = []
for i in range(len(result_files)):
combined_sizes.append(mean_sizes[i])
combined_labels.append('%s (Mean)' % labels[i])
combined_sizes.append(median_sizes[i])
combined_labels.append('%s (Median)' % labels[i])
myplot.cdf(combined_sizes, height_scale=0.7,
xlabel='Total Data Exchanged [MB]', labels=combined_labels,
xlim=(0, 5), labelspacing=0.1, handletextpad=0.4,
filename=os.path.join(args.outdir, 'size.pdf'))
def plot_browser(machine, remote, result_files):
##
## plot 1: compare 4-slice version of all protocols
##
out_filename, out_filepath = outfile(args.opt, remote, machine,\
extra_tag='_proto-comparison')
# need to make alternate dict of result paths
temp_result_files = {}
if 'spp_four-slices' in result_files:
temp_result_files['spp'] = result_files['spp_four-slices']
if 'spp_mod_four-slices' in result_files:
temp_result_files['spp_mod'] = result_files['spp_mod_four-slices']
if 'ssl_four-slices' in result_files:
temp_result_files['ssl'] = result_files['ssl_four-slices']
if 'ssl_mod_four-slices' in result_files:
temp_result_files['ssl_mod'] = result_files['ssl_mod_four-slices']
if 'fwd_four-slices' in result_files:
temp_result_files['fwd'] = result_files['fwd_four-slices']
if 'fwd_mod_four-slices' in result_files:
temp_result_files['fwd_mod'] = result_files['fwd_mod_four-slices']
if 'pln_four-slices' in result_files:
temp_result_files['pln'] = result_files['pln_four-slices']
if 'pln_mod_four-slices' in result_files:
temp_result_files['pln_mod'] = result_files['pln_mod_four-slices']
ys = [] # holds arrays of y values, 1 per series
labels = []
plot_title = ''
for protocol in PROTOCOLS[args.opt]:
if protocol not in temp_result_files: continue
filepath = temp_result_files[protocol]
print '[IN]', protocol, filepath
data = numpy.loadtxt(filepath)
if len(data) == 0 or\
len(data.shape) != 2 or\
data.shape[1] != 3:
print 'WARNING: malformed data: %s' % filepath
continue
transform = numpy.vectorize(DATA_TRANSFORMS[args.opt])
print protocol, max(transform(data[:, 2]))
ys.append(transform(data[:, 2]))
if protocol == 'spp':
labels.append(LEGEND_STRINGS[protocol] + ' (4 Ctx)')
elif protocol == 'spp_mod':
labels.append('mcTLS (4 Ctx, Nagle Off)')
else:
labels.append(LEGEND_STRINGS[protocol])
plot_title = title(args.opt, remote, data)
print '[OUT]', out_filepath
myplot.cdf(ys, labels=labels, xlabel=X_AXIS[args.opt],\
#title=plot_title,\
#builds = [[], [3], [3, 1, 2], [3, 1, 2, 0], [3, 1, 2, 0, 4]],\
filename=out_filepath, **MANUAL_ARGS[out_filename])
##
## plot 2: compare slice strategies (SPP only)
##
out_filename, out_filepath = outfile(args.opt, remote, machine,\
extra_tag='_slicing-comparison')
ys = [] # holds arrays of y values, 1 per series
labels = []
plot_title = ''
SLICE_LEGEND = {
'spp_one-slice': 'mcTLS (1 Ctx)',
'spp_mod_one-slice': 'mcTLS (1 Ctx, Nagle Off)',
'spp_four-slices': 'mcTLS (4 Ctx)',
'spp_mod_four-slices': 'mcTLS (4 Ctx, Nagle Off)',
'spp_slice-per-header': 'mcTLS (Ctx per Hdr)',
'spp_mod_slice-per-header': 'mcTLS (Ctx per Hdr, Nagle Off)',
}
for protocol in ('spp_one-slice', 'spp_mod_one-slice', 'spp_four-slices',
'spp_mod_four-slices', 'spp_slice-per-header',
'spp_mod_slice-per-header'):
if protocol not in result_files: continue
filepath = result_files[protocol]
print '[IN]', protocol, filepath
data = numpy.loadtxt(filepath)
if len(data) == 0 or\
len(data.shape) != 2 or\
data.shape[1] != 3:
print 'WARNING: malformed data: %s' % filepath
continue
transform = numpy.vectorize(DATA_TRANSFORMS[args.opt])
ys.append(transform(data[:, 2]))
labels.append(SLICE_LEGEND[protocol])
plot_title = title(args.opt, remote, data)
print '[OUT]', out_filepath
myplot.cdf(ys, labels=labels, xlabel=X_AXIS[args.opt],\
#title=plot_title,\
filename=out_filepath, **MANUAL_ARGS[out_filename])
def plot_browser(machine, remote, result_files):
##
## plot 1: compare 4-slice version of all protocols
##
out_filename, out_filepath = outfile(args.opt, remote, machine,\
extra_tag='_proto-comparison')
# need to make alternate dict of result paths
temp_result_files = {}
if 'spp_four-slices' in result_files:
temp_result_files['spp'] = result_files['spp_four-slices']
if 'spp_mod_four-slices' in result_files:
temp_result_files['spp_mod'] = result_files['spp_mod_four-slices']
if 'ssl_four-slices' in result_files:
temp_result_files['ssl'] = result_files['ssl_four-slices']
if 'ssl_mod_four-slices' in result_files:
temp_result_files['ssl_mod'] = result_files['ssl_mod_four-slices']
if 'fwd_four-slices' in result_files:
temp_result_files['fwd'] = result_files['fwd_four-slices']
if 'fwd_mod_four-slices' in result_files:
temp_result_files['fwd_mod'] = result_files['fwd_mod_four-slices']
if 'pln_four-slices' in result_files:
temp_result_files['pln'] = result_files['pln_four-slices']
if 'pln_mod_four-slices' in result_files:
temp_result_files['pln_mod'] = result_files['pln_mod_four-slices']
ys = [] # holds arrays of y values, 1 per series
labels = []
plot_title = ''
for protocol in PROTOCOLS[args.opt]:
if protocol not in temp_result_files: continue
filepath = temp_result_files[protocol]
print '[IN]', protocol, filepath
data = numpy.loadtxt(filepath)
if len(data) == 0 or\
len(data.shape) != 2 or\
data.shape[1] != 3:
print 'WARNING: malformed data: %s' % filepath
continue
transform = numpy.vectorize(DATA_TRANSFORMS[args.opt])
print protocol, max(transform(data[:,2]))
ys.append(transform(data[:,2]))
if protocol == 'spp':
labels.append(LEGEND_STRINGS[protocol] + ' (4 Ctx)')
elif protocol == 'spp_mod':
labels.append('mcTLS (4 Ctx, Nagle Off)')
else:
labels.append(LEGEND_STRINGS[protocol])
plot_title = title(args.opt, remote, data)
print '[OUT]', out_filepath
myplot.cdf(ys, labels=labels, xlabel=X_AXIS[args.opt],\
#title=plot_title,\
#builds = [[], [3], [3, 1, 2], [3, 1, 2, 0], [3, 1, 2, 0, 4]],\
filename=out_filepath, **MANUAL_ARGS[out_filename])
##
## plot 2: compare slice strategies (SPP only)
##
out_filename, out_filepath = outfile(args.opt, remote, machine,\
extra_tag='_slicing-comparison')
ys = [] # holds arrays of y values, 1 per series
labels = []
plot_title = ''
SLICE_LEGEND = {
'spp_one-slice': 'mcTLS (1 Ctx)',
'spp_mod_one-slice': 'mcTLS (1 Ctx, Nagle Off)',
'spp_four-slices': 'mcTLS (4 Ctx)',
'spp_mod_four-slices': 'mcTLS (4 Ctx, Nagle Off)',
'spp_slice-per-header': 'mcTLS (Ctx per Hdr)',
'spp_mod_slice-per-header': 'mcTLS (Ctx per Hdr, Nagle Off)',
}
for protocol in ('spp_one-slice', 'spp_mod_one-slice', 'spp_four-slices', 'spp_mod_four-slices', 'spp_slice-per-header', 'spp_mod_slice-per-header'):
if protocol not in result_files: continue
filepath = result_files[protocol]
print '[IN]', protocol, filepath
data = numpy.loadtxt(filepath)
if len(data) == 0 or\
len(data.shape) != 2 or\
data.shape[1] != 3:
print 'WARNING: malformed data: %s' % filepath
continue
transform = numpy.vectorize(DATA_TRANSFORMS[args.opt])
ys.append(transform(data[:,2]))
labels.append(SLICE_LEGEND[protocol])
plot_title = title(args.opt, remote, data)
print '[OUT]', out_filepath
myplot.cdf(ys, labels=labels, xlabel=X_AXIS[args.opt],\
#title=plot_title,\
filename=out_filepath, **MANUAL_ARGS[out_filename])