Stores per-iteration bench values in BenchDataPoint.

bench/bench_util.py

 '''
 Created on May 19, 2011
 
 @author: bungeman
 '''
 
 import re
 import math
 
 # bench representation algorithm constant names
@@ skipping 21 matching lines @@
 SETTINGS_RE_COMPILED = re.compile(SETTINGS_RE)
 BENCH_RE_COMPILED = re.compile(BENCH_RE)
 TIME_RE_COMPILED = re.compile(TIME_RE)
 CONFIG_RE_COMPILED = re.compile(CONFIG_RE)
 TILE_RE_COMPILED = re.compile(TILE_RE)
 TILE_LAYOUT_RE_COMPILED = re.compile(TILE_LAYOUT_RE)
 
 class BenchDataPoint:
     """A single data point produced by bench.
 
-    (str, str, str, float, {str:str}, str, [floats])"""
+    (str, str, str, float, {str:str}, str, [floats], [floats])"""

[borenet, 2014/03/12 15:54:32]

I'd really prefer to have these parameters documented per-argument with some
explanation.

[benchen, 2014/03/12 16:22:51]

They are member variables. I added explanations in __init__. Sorry for the
historical confusion.

[borenet, 2014/03/12 17:13:06]

Thanks.  This is okay, but I'd still prefer the documentation to be in the
docstring for __init__, like this:

def __init__(....):
  """Instantiate a BenchDataPoint.

  Args:
    bench: string; name of the benchmark to measure
    .... etc ....
  """

     def __init__(self, bench, config, time_type, time, settings,
-                 tile_layout='', per_tile_values=[]):
+                 tile_layout='', per_tile_values=[], per_iter_time=[]):
         self.bench = bench
         self.config = config
         self.time_type = time_type
         self.time = time
         self.settings = settings
         # how tiles cover the whole picture. '5x3' means 5 columns and 3 rows.
         self.tile_layout = tile_layout
         # list of per_tile bench values, if applicable
         self.per_tile_values = per_tile_values
+        # list of per-iteration bench time
+        self.per_iter_time = per_iter_time
 
     def __repr__(self):
         return "BenchDataPoint(%s, %s, %s, %s, %s)" % (
                    str(self.bench),
                    str(self.config),
                    str(self.time_type),
                    str(self.time),
                    str(self.settings),
                )
 
@@ skipping 38 matching lines @@
                     x = int(round(0.25 * self._len + 0.5))
                 else:
                     raise Exception("invalid representation algorithm %s!" %
                                     representation)
                 self._rep = self._data[x - 1]
 
     def compute(self):
         return self._rep
 
 def _ParseAndStoreTimes(config_re_compiled, is_per_tile, line, bench,
-                        value_dic, layout_dic, representation=None):
+                        value_dic, layout_dic):
     """Parses given bench time line with regex and adds data to value_dic.
 
     config_re_compiled: precompiled regular expression for parsing the config
         line.
     is_per_tile: boolean indicating whether this is a per-tile bench.
         If so, we add tile layout into layout_dic as well.
     line: input string line to parse.
     bench: name of bench for the time values.
     value_dic: dictionary to store bench values. See bench_dic in parse() below.
     layout_dic: dictionary to store tile layouts. See parse() for descriptions.
-    representation: should match one of the ALGORITHM_XXX types."""
+    """
 
     for config in config_re_compiled.finditer(line):
         current_config = config.group(1)
         tile_layout = ''
         if is_per_tile:  # per-tile bench, add name prefix
             current_config = 'tile_' + current_config
             layouts = TILE_LAYOUT_RE_COMPILED.search(line)
             if layouts and len(layouts.groups()) == 2:
               tile_layout = '%sx%s' % layouts.groups()
         times = config.group(2)
         for new_time in TIME_RE_COMPILED.finditer(times):
             current_time_type = new_time.group(1)
             iters = [float(i) for i in
                      new_time.group(2).strip().split(',')]
             value_dic.setdefault(bench, {}).setdefault(
                 current_config, {}).setdefault(current_time_type, []).append(
-                    _ListAlgorithm(iters, representation).compute())
+                    iters)
             layout_dic.setdefault(bench, {}).setdefault(
                 current_config, {}).setdefault(current_time_type, tile_layout)
 
 # TODO(bensong): switch to reading JSON output when available. This way we don't
 # need the RE complexities.
 def parse(settings, lines, representation=None):
     """Parses bench output into a useful data structure.
 
     ({str:str}, __iter__ -> str) -> [BenchDataPoint]
     representation is one of the ALGORITHM_XXX types."""
 
     benches = []
     current_bench = None
-    bench_dic = {}  # [bench][config][time_type] -> [list of bench values]
+    # [bench][config][time_type] -> [[per-iter values]] where per-tile config
+    # has per-iter value list for each tile [[<tile1_iter1>,<tile1_iter2>,...],
+    # [<tile2_iter1>,<tile2_iter2>,...],...], while non-per-tile config only
+    # contains one list of iterations [[iter1, iter2, ...]].
+    bench_dic = {}
     # [bench][config][time_type] -> tile_layout
     layout_dic = {}
 
     for line in lines:
 
         # see if this line is a settings line
         settingsMatch = SETTINGS_RE_COMPILED.search(line)
         if (settingsMatch):
             settings = dict(settings)
             for settingMatch in PER_SETTING_RE_COMPILED.finditer(settingsMatch.group(1)):
                 if (settingMatch.group(2)):
                     settings[settingMatch.group(1)] = settingMatch.group(2)
                 else:
                     settings[settingMatch.group(1)] = True
 
         # see if this line starts a new bench
         new_bench = BENCH_RE_COMPILED.search(line)
         if new_bench:
             current_bench = new_bench.group(1)
 
         # add configs on this line to the bench_dic
         if current_bench:
             if line.startswith('  tile_') :
                 _ParseAndStoreTimes(TILE_RE_COMPILED, True, line, current_bench,
-                                    bench_dic, layout_dic, representation)
+                                    bench_dic, layout_dic)
             else:
                 _ParseAndStoreTimes(CONFIG_RE_COMPILED, False, line,
-                                    current_bench,
-                                    bench_dic, layout_dic, representation)
+                                    current_bench, bench_dic, layout_dic)
 
-    # append benches to list, use the total time as final bench value.
+    # append benches to list
     for bench in bench_dic:
         for config in bench_dic[bench]:
             for time_type in bench_dic[bench][config]:
                 tile_layout = ''
-                per_tile_values = []
+                per_tile_values = []  # empty for non-per-tile configs
+                per_iter_time = []  # empty for per-tile configs
+                bench_summary = None  # a single final bench value
                 if len(bench_dic[bench][config][time_type]) > 1:
-                    # per-tile values, extract tile_layout
-                    per_tile_values = bench_dic[bench][config][time_type]
+                    # per-tile config; compute representation for each tile
+                    per_tile_values = [
+                        _ListAlgorithm(iters, representation).compute()
+                            for iters in bench_dic[bench][config][time_type]]
+                    # use sum of each tile representation for total bench value
+                    bench_summary = sum(per_tile_values)
+                    # extract tile layout
                     tile_layout = layout_dic[bench][config][time_type]
+                else:
+                    # get the list of per-iteration values
+                    per_iter_time = bench_dic[bench][config][time_type][0]
+                    bench_summary = _ListAlgorithm(
+                        per_iter_time, representation).compute()
                 benches.append(BenchDataPoint(
                     bench,
                     config,
                     time_type,
-                    sum(bench_dic[bench][config][time_type]),
+                    bench_summary,
                     settings,
                     tile_layout,
-                    per_tile_values))
+                    per_tile_values,
+                    per_iter_time))
 
     return benches
 
 class LinearRegression:
     """Linear regression data based on a set of data points.
 
     ([(Number,Number)])
     There must be at least two points for this to make sense."""
     def __init__(self, points):
         n = len(points)
@@ skipping 78 matching lines @@
     """
     return '<a href="http://code.google.com/p/skia/source/detail?r=%s">%s</a>'%(
         revision_number, revision_number)
 
 def main():
     foo = [[0.0, 0.0], [0.0, 1.0], [0.0, 2.0], [0.0, 3.0]]
     LinearRegression(foo)
 
 if __name__ == "__main__":
     main()