Update loading_measurement_analyzer to display CPU vs Network breakdowns.

tools/perf/measurements/loading_measurement_analyzer.py

Index: tools/perf/measurements/loading_measurement_analyzer.py
diff --git a/tools/perf/measurements/loading_measurement_analyzer.py b/tools/perf/measurements/loading_measurement_analyzer.py
index a5abe11ac685340ccca200c0a57dfb381d34a46b..395b61d2558dfdc819dface785fe73b4c13164ac 100755
--- a/tools/perf/measurements/loading_measurement_analyzer.py
+++ b/tools/perf/measurements/loading_measurement_analyzer.py
@@ -29,6 +29,9 @@ class LoadingMeasurementAnalyzer(object):
     self.totals = collections.defaultdict(list)
     self.maxes = collections.defaultdict(list)
     self.avgs = collections.defaultdict(list)
+    self.load_times = []
+    self.cpu_times = []
+    self.network_percents = []
     self.num_rows_parsed = 0
     self.num_slowest_urls = options.num_slowest_urls
     if options.rank_csv_file:
@@ -42,17 +45,30 @@ class LoadingMeasurementAnalyzer(object):
         if (options.rank_limit and
             self._GetRank(row['url']) > options.rank_limit):
           continue
+        cpu_time = 0
+        load_time = float(row['load_time (ms)'])
+        if load_time < 0:
+          print 'Skipping %s due to negative load time' % row['url']
+          continue
         for key, value in row.iteritems():
-          if key in ('url', 'dom_content_loaded_time (ms)', 'load_time (ms)'):
+          if key in ('url', 'load_time (ms)', 'dom_content_loaded_time (ms)'):
             continue
           if not value or value == '-':
             continue
+          value = float(value)
           if '_avg' in key:
-            self.avgs[key].append((float(value), row['url']))
+            self.avgs[key].append((value, row['url']))
           elif '_max' in key:
-            self.maxes[key].append((float(value), row['url']))
+            self.maxes[key].append((value, row['url']))
           else:
-            self.totals[key].append((float(value), row['url']))
+            self.totals[key].append((value, row['url']))

[nduca, 2013/07/22 23:37:00]

so you define network time as the unaccounted-for time?

what happens when you have this?
  Begin ts=10 Foo
  Begin ts=15 Bar
  End    ts=20 Bar
  End    ts=25 Foo

This gives us foo duration=15 Bar duration=5. Cpu time consumed is actually 15
in this case...

+            cpu_time += value
+        self.load_times.append((load_time, row['url']))
+        self.cpu_times.append((cpu_time, row['url']))
+        if options.show_network:
+          network_time = load_time - cpu_time
+          self.totals['Network (ms)'].append((network_time, row['url']))
+          self.network_percents.append((network_time / load_time, row['url']))
         self.num_rows_parsed += 1
         if options.max_rows and self.num_rows_parsed == int(options.max_rows):
           break
@@ -73,17 +89,23 @@ class LoadingMeasurementAnalyzer(object):
     sum_totals = {}
     for key, values in self.totals.iteritems():
       sum_totals[key] = sum([v[0] for v in values])
-    total_time = sum(sum_totals.values())
+    total_cpu_time = sum([v[0] for v in self.cpu_times])
+    total_page_load_time = sum([v[0] for v in self.load_times])
 
     print
     print 'Total URLs: ', self.num_rows_parsed
-    print 'Total time: %ds' % int(round(total_time / 1000))
+    print 'Total CPU time: %ds' % int(round(total_cpu_time / 1000))
+    print 'Total page load time: %ds' % int(round(total_page_load_time / 1000))
+    print 'Average CPU time: %dms' % int(round(
+        total_cpu_time / self.num_rows_parsed))
+    print 'Average page load time: %dms' % int(round(
+        total_page_load_time / self.num_rows_parsed))
     print
     for key, value in sorted(sum_totals.iteritems(), reverse=True,
                              key=lambda i: i[1]):
       output_key = '%30s: ' % key.replace(' (ms)', '')
       output_value = '%10ds ' % (value / 1000)
-      output_percent = '%.1f%%' % (100 * value / total_time)
+      output_percent = '%.1f%%' % (100 * value / total_page_load_time)
       print output_key, output_value, output_percent
 
     if not self.num_slowest_urls:
@@ -98,6 +120,15 @@ class LoadingMeasurementAnalyzer(object):
       for value, url in slowest:
         print '\t', '%dms\t' % value, url, '(#%s)' % self._GetRank(url)
 
+    if self.network_percents:
+      print
+      print 'Top %d highest network to CPU time ratios:' % self.num_slowest_urls
+      for percent, url in sorted(
+          self.network_percents, reverse=True)[:self.num_slowest_urls]:
+        percent *= 100
+        print '\t', '%.1f%%' % percent, url, '(#%s)' % self._GetRank(url)
+
+
 def main(argv):
   prog_desc = 'Parses CSV output from the loading_measurement'
   parser = optparse.OptionParser(usage=('%prog [options]' + '\n\n' + prog_desc))
@@ -109,6 +140,8 @@ def main(argv):
   parser.add_option('--rank-csv-file', help='A CSV file of <rank,url>')
   parser.add_option('--rank-limit', type='int',
                     help='Only process pages higher than this rank')
+  parser.add_option('--show-network', action='store_true',
+                    help='Whether to display Network as a category')
 
   options, args = parser.parse_args(argv[1:])