New plot script: Run duration over time

requirements.txt

@@ -1,5 +1,7 @@
+matplotlib>=2.1.1
+numpy==1.16.4
+pandas==0.24.2
+pexpect>=4.7.0
 scipy>=1.0
 termcolor>=1.0
-terminaltables>=3.0
-matplotlib>=2.1.1
-pexpect>=4.7.0
+terminaltables>=3.0

scripts/plot_run_duration_over_time.py

@@ -0,0 +1,85 @@
+#!/usr/bin/env python3
+
+# Takes a single benchmark output JSON and plots the duration of item executions over time
+
+import json
+import matplotlib.patches as mpatches
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+import sys
+
+if(len(sys.argv) != 2):
+    print('Usage: ' + sys.argv[0] + ' benchmark.json')
+    exit()
+
+with open(sys.argv[1]) as json_file:
+    data_json = json.load(json_file)['benchmarks']
+
+# Build flat list of {name, begin, duration} entries for every benchmark item run
+data = []
+benchmark_names = []
+for benchmark_json in data_json:
+	name = benchmark_json['name']
+	benchmark_names.append(name)
+	for run_json in benchmark_json['successful_runs']:
+		data.append({'name': name, 'begin': run_json['begin'] / 1e9, 'duration': run_json['duration'], 'success': True})
+	for run_json in benchmark_json['unsuccessful_runs']:
+		data.append({'name': name, 'begin': run_json['begin'] / 1e9, 'duration': run_json['duration'], 'success': False})
+
+
+df = pd.DataFrame(data).reset_index()
+
+# Set the colors
+benchmark_names.sort()  # Sort the benchmarks for a deterministic color mapping
+name_to_color = {}
+prop_cycle = plt.rcParams['axes.prop_cycle']
+default_colors = prop_cycle.by_key()['color']
+
+fig, ax = plt.subplots()
+for i in range(0, len(benchmark_names)):
+	benchmark_name = benchmark_names[i]
+	color = default_colors[i % len(default_colors)]
+	name_to_color[benchmark_name] = color
+
+	filtered_df = df[df['name'] == benchmark_name]
+
+	# Plot into combined graph
+	for success in [True, False]:
+		if not filtered_df[filtered_df['success']==success].empty:
+			filtered_df[filtered_df['success']==success].plot(ax=ax, kind='scatter', x='begin', y='duration', c=color, figsize=(12, 9), s=5, marker=('o' if success else 'x'), linewidth=1)
+
+# Finish combined graph
+ax.set_xlabel('Seconds since start')
+ax.set_ylabel('Execution duration [ns]')
+ax.grid(True, alpha=.3)
+
+# Add legend to combined graph
+handles = []
+for name, color in name_to_color.items():
+	handles.append(mpatches.Patch(label=name, color=color))
+plt.legend(handles=handles)
+
+# Write combined graph to file
+basename = sys.argv[1].replace('.json', '')
+plt.savefig(basename + '.pdf')
+plt.savefig(basename + '.png')
+
+# Plot detailed graph
+grouped_df = df.reset_index().groupby('name')
+fig, axes = plt.subplots(nrows=len(benchmark_names), ncols=1, figsize=(12, 3*len(benchmark_names)), sharex=True)
+for (key, ax) in zip(grouped_df.groups.keys(), axes.flatten()):
+	filtered_df = grouped_df.get_group(key)
+	for success in [True, False]:
+		if not filtered_df[filtered_df['success']==success].empty:
+			filtered_df[filtered_df['success']==success].plot(ax=ax, kind='scatter', x='begin', y='duration', c=name_to_color[key], s=5, marker=('o' if success else 'x'), linewidth=1)
+	ax.set_title(key)
+	ax.set_xlabel('Seconds since start')
+	ax.set_ylabel('Execution duration [ns]')
+	ax.yaxis.set_label_coords(-.08, 0.5)
+	ax.grid(True, alpha=.3)
+fig.tight_layout()
+
+# Write detailed graph to file
+plt.savefig(basename + '_detailed.pdf')
+plt.savefig(basename + '_detailed.png')