Improve analysis scripts

Author: gjbex

source-code/mpi4py/analyze_mpifitness_data.py

@@ -7,26 +7,33 @@
 
 
 def accumulate(data_file):
-    regex = re.compile(r'^([a-z]+).+?:\s+(\d+(?:\.\d+)(?:e[+-]\d+)?)')
+    regex = re.compile(r'^([a-z]+).+?(\d+\.\d+):\s+(\d+(?:\.\d+)(?:e[+-]\d+)?)')
     timings = dict()
+    times = dict()
     for line in data_file:
         match = regex.search(line)
         if match:
             test = match.group(1)
             time = float(match.group(2))
+            duration = float(match.group(3))
             if test not in timings:
                 timings[test] = list()
-            timings[test].append(time)
-    return timings
+                times[test] = list()
+            timings[test].append(duration)
+            times[test].append(time)
+    data = dict()
+    for test in timings:
+        data[test] = np.array([times[test], timings[test]]).T
+    return data
 
 
-def print_stats(test, data_list):
+def print_stats(test, data):
     print(f'{test}:')
-    data = np.array(data_list)
     print(f'    min:    {data.min()}')
     print(f'    median: {np.median(data)}')
     print(f'    mean:   {data.mean()}')
     print(f'    max:    {data.max()}')
+    print(f'    stddev: {np.std(data)}')
     print(f'    n:      {len(data)}')
 
 
@@ -36,7 +43,7 @@ def main():
     options = arg_parser.parse_args()
     timings = accumulate(options.file)
     for test, data_list in timings.items():
-        print_stats(test, data_list)
+        print_stats(test, data_list[:, 1])
     return 0
 
 

source-code/mpi4py/plot_time_distribution.py

@@ -21,10 +21,11 @@ def main():
     arg_parser.add_argument('--log', action='store_true', help='use log x-axis')
     options = arg_parser.parse_args()
     timings = accumulate(options.file)
+    data = timings[options.test][:, 1]
     if options.rug:
-        grid = sns.distplot(timings[options.test], rug=True, hist=False)
+        grid = sns.distplot(data, rug=True, hist=False)
     else:
-        grid = sns.distplot(timings[options.test], bins=options.bins)
+        grid = sns.distplot(data, bins=options.bins)
     if options.log:
         grid.set(xscale='log')
     grid.set(title=options.test)

source-code/mpi4py/plot_timeline.py

@@ -0,0 +1,31 @@
+#!/usr/bin/env python
+
+from argparse import ArgumentParser, FileType
+import matplotlib.pyplot as plt
+import seaborn as sns
+import sys
+from analyze_mpifitness_data import accumulate
+
+
+def main():
+    arg_parser = ArgumentParser(description='plot MPI time distirbution')
+    arg_parser.add_argument('--file', required=True, type=FileType('r'),
+                            help='file to plot data from')
+    arg_parser.add_argument('--test', required=True,
+                            choices=['pingpong', 'bcast', 'scatter', 'gather',
+                                     'alltoall', 'reduce'],
+                            help='test to visualize')
+    arg_parser.add_argument('--log', action='store_true', help='use log x-axis')
+    options = arg_parser.parse_args()
+    timings = accumulate(options.file)
+    data = timings[options.test]
+    grid = sns.scatterplot(x=(data[:, 0] - data[0, 0]), y=data[:, 1], alpha=0.6)
+    if options.log:
+        grid.set(yscale='log')
+    grid.set(title=options.test)
+    plt.show()
+    return 0
+
+
+if __name__ == '__main__':
+    sys.exit(main())