russell@unturf.com
88a3a6718c
- Add scripts/generate-perf-charts.py for matplotlib visualizations - Update generate-perf-report.sh for versioned directories (reports/TAG/) - Add make perf-charts and make perf-all targets - Add GitLab CI perf-report job to auto-commit after tagged releases - Generated 6 charts for 4.2.0: dashboard, duration, histogram, leaders, etc. Each release now gets its own reports/TAG/ directory with: - perf.json (raw timing data) - perf.md (markdown report) - chart-*.png (6 visualizations)
359 lines
14 KiB
Python
Executable file
359 lines
14 KiB
Python
Executable file
#!/usr/bin/env python3
|
|
"""
|
|
Generate performance visualization charts from CI timing data.
|
|
Outputs PNG charts to /tmp/artifacts/ for collection by unsandbox.
|
|
|
|
Usage: python3 generate-perf-charts.py [perf.json]
|
|
Via unsandbox: build/un -a -f reports/TAG/perf.json scripts/generate-perf-charts.py
|
|
"""
|
|
|
|
import json
|
|
import sys
|
|
import os
|
|
|
|
# Create artifacts directory (unsandbox collects from /tmp/artifacts/)
|
|
os.makedirs('/tmp/artifacts', exist_ok=True)
|
|
|
|
# Load data from file or stdin
|
|
# When run via unsandbox, files are uploaded to /tmp/input/
|
|
possible_paths = [
|
|
'/tmp/input/perf.json', # Via unsandbox -f
|
|
'/tmp/input/perf-4.2.0.json', # Legacy format
|
|
]
|
|
|
|
if len(sys.argv) > 1:
|
|
possible_paths.insert(0, sys.argv[1])
|
|
|
|
# Also try to find any JSON file in /tmp/input/
|
|
if os.path.exists('/tmp/input'):
|
|
for f in os.listdir('/tmp/input'):
|
|
if f.endswith('.json'):
|
|
possible_paths.append(f'/tmp/input/{f}')
|
|
|
|
data = None
|
|
for path in possible_paths:
|
|
try:
|
|
with open(path) as f:
|
|
data = json.load(f)
|
|
print(f'Loaded data from {path}')
|
|
break
|
|
except (FileNotFoundError, json.JSONDecodeError):
|
|
continue
|
|
|
|
if data is None:
|
|
# Try stdin as last resort
|
|
try:
|
|
data = json.load(sys.stdin)
|
|
print('Loaded data from stdin')
|
|
except:
|
|
print('Error: Could not load performance data')
|
|
print(f'Tried: {possible_paths}')
|
|
sys.exit(1)
|
|
|
|
# Import matplotlib (available in unsandbox)
|
|
import matplotlib.pyplot as plt
|
|
import matplotlib.patches as mpatches
|
|
import numpy as np
|
|
|
|
# Set style
|
|
plt.style.use('seaborn-v0_8-darkgrid' if 'seaborn-v0_8-darkgrid' in plt.style.available else 'ggplot')
|
|
plt.rcParams['figure.facecolor'] = '#1a1a2e'
|
|
plt.rcParams['axes.facecolor'] = '#16213e'
|
|
plt.rcParams['text.color'] = '#eee'
|
|
plt.rcParams['axes.labelcolor'] = '#eee'
|
|
plt.rcParams['xtick.color'] = '#eee'
|
|
plt.rcParams['ytick.color'] = '#eee'
|
|
plt.rcParams['axes.edgecolor'] = '#444'
|
|
plt.rcParams['grid.color'] = '#333'
|
|
plt.rcParams['font.size'] = 10
|
|
|
|
# Extract data
|
|
languages = data['languages']
|
|
tag = data['tag']
|
|
summary = data['summary']
|
|
|
|
# Sort by duration
|
|
langs_sorted = sorted(languages, key=lambda x: x['duration_seconds'], reverse=True)
|
|
names = [l['language'] for l in langs_sorted]
|
|
durations = [l['duration_seconds'] for l in langs_sorted]
|
|
queued = [l['queued_duration'] for l in langs_sorted]
|
|
|
|
# Language categories
|
|
COMPILED = {'c', 'cpp', 'go', 'rust', 'java', 'kotlin', 'swift', 'csharp', 'fsharp',
|
|
'haskell', 'ocaml', 'd', 'nim', 'zig', 'crystal', 'fortran', 'cobol',
|
|
'objc', 'v', 'dart'}
|
|
INTERPRETED = {'python', 'javascript', 'typescript', 'ruby', 'php', 'perl', 'lua',
|
|
'bash', 'r', 'awk', 'tcl', 'scheme', 'commonlisp', 'clojure', 'elixir',
|
|
'erlang', 'groovy', 'raku', 'julia', 'prolog', 'forth', 'powershell', 'deno'}
|
|
|
|
categories = []
|
|
for l in langs_sorted:
|
|
lang = l['language']
|
|
if lang in COMPILED:
|
|
categories.append('compiled')
|
|
elif lang in INTERPRETED:
|
|
categories.append('interpreted')
|
|
else:
|
|
categories.append('other')
|
|
|
|
# Color palette
|
|
colors_by_cat = {'compiled': '#e94560', 'interpreted': '#0f3460', 'other': '#533483'}
|
|
colors = [colors_by_cat.get(c, '#666') for c in categories]
|
|
|
|
# ============================================================================
|
|
# Chart 1: Horizontal Bar Chart - Duration by Language
|
|
# ============================================================================
|
|
fig, ax = plt.subplots(figsize=(12, 14))
|
|
|
|
y_pos = np.arange(len(names))
|
|
bars = ax.barh(y_pos, durations, color=colors, edgecolor='#fff', linewidth=0.5)
|
|
|
|
# Add value labels
|
|
for i, (bar, dur) in enumerate(zip(bars, durations)):
|
|
ax.text(dur + 1, bar.get_y() + bar.get_height()/2, f'{dur}s',
|
|
va='center', ha='left', fontsize=8, color='#aaa')
|
|
|
|
ax.set_yticks(y_pos)
|
|
ax.set_yticklabels(names, fontsize=9)
|
|
ax.invert_yaxis()
|
|
ax.set_xlabel('Duration (seconds)', fontsize=12)
|
|
ax.set_title(f'UN Inception v{tag} - Test Duration by Language\n42 Languages • 638 Tests • 100% Pass Rate',
|
|
fontsize=14, fontweight='bold', color='#fff', pad=20)
|
|
|
|
# Legend
|
|
compiled_patch = mpatches.Patch(color='#e94560', label='Compiled')
|
|
interpreted_patch = mpatches.Patch(color='#0f3460', label='Interpreted')
|
|
ax.legend(handles=[compiled_patch, interpreted_patch], loc='lower right',
|
|
facecolor='#1a1a2e', edgecolor='#444', labelcolor='#eee')
|
|
|
|
# Stats annotation
|
|
stats_text = f"Avg: {summary['avg_duration_seconds']}s | Min: {summary['min_duration_seconds']}s | Max: {summary['max_duration_seconds']}s"
|
|
ax.text(0.5, -0.05, stats_text, transform=ax.transAxes, ha='center', fontsize=10, color='#888')
|
|
|
|
plt.tight_layout()
|
|
plt.savefig('/tmp/artifacts/chart-duration-by-language.png', dpi=150, facecolor='#1a1a2e')
|
|
print('✓ chart-duration-by-language.png')
|
|
|
|
# ============================================================================
|
|
# Chart 2: Category Comparison (Compiled vs Interpreted)
|
|
# ============================================================================
|
|
fig, axes = plt.subplots(1, 2, figsize=(14, 6))
|
|
|
|
# Pie chart
|
|
compiled_count = sum(1 for c in categories if c == 'compiled')
|
|
interpreted_count = sum(1 for c in categories if c == 'interpreted')
|
|
|
|
ax1 = axes[0]
|
|
sizes = [compiled_count, interpreted_count]
|
|
labels = [f'Compiled\n({compiled_count})', f'Interpreted\n({interpreted_count})']
|
|
explode = (0.02, 0.02)
|
|
wedges, texts, autotexts = ax1.pie(sizes, labels=labels, autopct='%1.0f%%',
|
|
colors=['#e94560', '#0f3460'],
|
|
explode=explode, startangle=90,
|
|
textprops={'color': '#eee', 'fontsize': 11},
|
|
wedgeprops={'edgecolor': '#fff', 'linewidth': 1})
|
|
ax1.set_title('Language Distribution', fontsize=12, color='#fff', fontweight='bold')
|
|
|
|
# Box plot comparing durations
|
|
ax2 = axes[1]
|
|
compiled_durations = [l['duration_seconds'] for l, c in zip(langs_sorted, categories) if c == 'compiled']
|
|
interpreted_durations = [l['duration_seconds'] for l, c in zip(langs_sorted, categories) if c == 'interpreted']
|
|
|
|
bp = ax2.boxplot([compiled_durations, interpreted_durations],
|
|
labels=['Compiled', 'Interpreted'],
|
|
patch_artist=True,
|
|
medianprops={'color': '#fff', 'linewidth': 2})
|
|
|
|
bp['boxes'][0].set_facecolor('#e94560')
|
|
bp['boxes'][1].set_facecolor('#0f3460')
|
|
for box in bp['boxes']:
|
|
box.set_edgecolor('#fff')
|
|
|
|
ax2.set_ylabel('Duration (seconds)', fontsize=11)
|
|
ax2.set_title('Duration Distribution by Category', fontsize=12, color='#fff', fontweight='bold')
|
|
|
|
# Add means as points
|
|
ax2.scatter([1], [np.mean(compiled_durations)], color='#fff', s=100, zorder=5, marker='D', label='Mean')
|
|
ax2.scatter([2], [np.mean(interpreted_durations)], color='#fff', s=100, zorder=5, marker='D')
|
|
|
|
plt.suptitle(f'UN Inception v{tag} - Compiled vs Interpreted', fontsize=14, fontweight='bold', color='#fff', y=1.02)
|
|
plt.tight_layout()
|
|
plt.savefig('/tmp/artifacts/chart-category-comparison.png', dpi=150, facecolor='#1a1a2e')
|
|
print('✓ chart-category-comparison.png')
|
|
|
|
# ============================================================================
|
|
# Chart 3: Duration Distribution Histogram with KDE-like curve
|
|
# ============================================================================
|
|
fig, ax = plt.subplots(figsize=(10, 6))
|
|
|
|
n, bins, patches = ax.hist(durations, bins=15, color='#e94560', edgecolor='#fff',
|
|
alpha=0.8, linewidth=1)
|
|
|
|
# Color gradient based on height
|
|
max_height = max(n)
|
|
for patch, height in zip(patches, n):
|
|
intensity = 0.3 + 0.7 * (height / max_height)
|
|
patch.set_facecolor(plt.cm.plasma(intensity))
|
|
|
|
ax.axvline(summary['avg_duration_seconds'], color='#00ff88', linestyle='--',
|
|
linewidth=2, label=f"Mean: {summary['avg_duration_seconds']}s")
|
|
ax.axvline(np.median(durations), color='#00d4ff', linestyle=':',
|
|
linewidth=2, label=f"Median: {int(np.median(durations))}s")
|
|
|
|
ax.set_xlabel('Duration (seconds)', fontsize=12)
|
|
ax.set_ylabel('Number of Languages', fontsize=12)
|
|
ax.set_title(f'UN Inception v{tag} - Test Duration Distribution\n42 Languages',
|
|
fontsize=14, fontweight='bold', color='#fff', pad=15)
|
|
ax.legend(facecolor='#1a1a2e', edgecolor='#444', labelcolor='#eee')
|
|
|
|
plt.tight_layout()
|
|
plt.savefig('/tmp/artifacts/chart-duration-histogram.png', dpi=150, facecolor='#1a1a2e')
|
|
print('✓ chart-duration-histogram.png')
|
|
|
|
# ============================================================================
|
|
# Chart 4: Top 10 Slowest vs Top 10 Fastest
|
|
# ============================================================================
|
|
fig, axes = plt.subplots(1, 2, figsize=(14, 6))
|
|
|
|
# Top 10 slowest
|
|
ax1 = axes[0]
|
|
slowest = langs_sorted[:10]
|
|
slow_names = [l['language'] for l in slowest]
|
|
slow_durations = [l['duration_seconds'] for l in slowest]
|
|
slow_colors = [colors_by_cat.get(c, '#666') for c in categories[:10]]
|
|
|
|
bars1 = ax1.barh(range(10), slow_durations, color=slow_colors, edgecolor='#fff')
|
|
ax1.set_yticks(range(10))
|
|
ax1.set_yticklabels(slow_names)
|
|
ax1.invert_yaxis()
|
|
ax1.set_xlabel('Duration (seconds)')
|
|
ax1.set_title('🐢 Top 10 Slowest', fontsize=12, color='#ff6b6b', fontweight='bold')
|
|
for i, (bar, dur) in enumerate(zip(bars1, slow_durations)):
|
|
ax1.text(dur + 1, i, f'{dur}s', va='center', fontsize=9, color='#aaa')
|
|
|
|
# Top 10 fastest
|
|
ax2 = axes[1]
|
|
fastest = langs_sorted[-10:][::-1]
|
|
fast_names = [l['language'] for l in fastest]
|
|
fast_durations = [l['duration_seconds'] for l in fastest]
|
|
fast_categories = [categories[names.index(n)] for n in fast_names]
|
|
fast_colors = [colors_by_cat.get(c, '#666') for c in fast_categories]
|
|
|
|
bars2 = ax2.barh(range(10), fast_durations, color=fast_colors, edgecolor='#fff')
|
|
ax2.set_yticks(range(10))
|
|
ax2.set_yticklabels(fast_names)
|
|
ax2.invert_yaxis()
|
|
ax2.set_xlabel('Duration (seconds)')
|
|
ax2.set_title('🚀 Top 10 Fastest', fontsize=12, color='#00ff88', fontweight='bold')
|
|
for i, (bar, dur) in enumerate(zip(bars2, fast_durations)):
|
|
ax2.text(dur + 1, i, f'{dur}s', va='center', fontsize=9, color='#aaa')
|
|
|
|
plt.suptitle(f'UN Inception v{tag} - Speed Leaders', fontsize=14, fontweight='bold', color='#fff', y=1.02)
|
|
plt.tight_layout()
|
|
plt.savefig('/tmp/artifacts/chart-speed-leaders.png', dpi=150, facecolor='#1a1a2e')
|
|
print('✓ chart-speed-leaders.png')
|
|
|
|
# ============================================================================
|
|
# Chart 5: Queue Time vs Execution Time Scatter
|
|
# ============================================================================
|
|
fig, ax = plt.subplots(figsize=(10, 8))
|
|
|
|
scatter_colors = [colors_by_cat.get(c, '#666') for c in categories]
|
|
scatter = ax.scatter(queued, durations, c=scatter_colors, s=100, alpha=0.8, edgecolors='#fff', linewidth=0.5)
|
|
|
|
# Label outliers
|
|
for i, (q, d, name) in enumerate(zip(queued, durations, names)):
|
|
if d > 70 or q > 200:
|
|
ax.annotate(name, (q, d), xytext=(5, 5), textcoords='offset points',
|
|
fontsize=8, color='#aaa')
|
|
|
|
ax.set_xlabel('Queue Duration (seconds)', fontsize=12)
|
|
ax.set_ylabel('Execution Duration (seconds)', fontsize=12)
|
|
ax.set_title(f'UN Inception v{tag} - Queue Time vs Execution Time\nEach point is one language',
|
|
fontsize=14, fontweight='bold', color='#fff', pad=15)
|
|
|
|
# Legend
|
|
compiled_patch = mpatches.Patch(color='#e94560', label='Compiled')
|
|
interpreted_patch = mpatches.Patch(color='#0f3460', label='Interpreted')
|
|
ax.legend(handles=[compiled_patch, interpreted_patch], loc='upper right',
|
|
facecolor='#1a1a2e', edgecolor='#444', labelcolor='#eee')
|
|
|
|
plt.tight_layout()
|
|
plt.savefig('/tmp/artifacts/chart-queue-vs-execution.png', dpi=150, facecolor='#1a1a2e')
|
|
print('✓ chart-queue-vs-execution.png')
|
|
|
|
# ============================================================================
|
|
# Chart 6: Summary Dashboard
|
|
# ============================================================================
|
|
fig = plt.figure(figsize=(16, 10))
|
|
|
|
# Create grid
|
|
gs = fig.add_gridspec(3, 3, hspace=0.4, wspace=0.3)
|
|
|
|
# Big title
|
|
fig.suptitle(f'UN Inception v{tag} Performance Dashboard', fontsize=20, fontweight='bold',
|
|
color='#fff', y=0.98)
|
|
|
|
# Metric cards (top row)
|
|
metrics = [
|
|
('Tests', str(summary['total_tests']), '#00ff88'),
|
|
('Languages', str(summary['total_languages']), '#00d4ff'),
|
|
('Pass Rate', '100%', '#ff6b6b'),
|
|
('Avg Time', f"{summary['avg_duration_seconds']}s", '#ffd93d'),
|
|
]
|
|
|
|
for i, (label, value, color) in enumerate(metrics):
|
|
ax = fig.add_subplot(gs[0, i] if i < 3 else gs[0, 2])
|
|
if i == 3:
|
|
ax = fig.add_axes([0.78, 0.75, 0.18, 0.15])
|
|
ax.set_xlim(0, 1)
|
|
ax.set_ylim(0, 1)
|
|
ax.text(0.5, 0.6, value, fontsize=32, fontweight='bold', color=color,
|
|
ha='center', va='center', transform=ax.transAxes)
|
|
ax.text(0.5, 0.2, label, fontsize=14, color='#888',
|
|
ha='center', va='center', transform=ax.transAxes)
|
|
ax.axis('off')
|
|
|
|
# Mini bar chart (bottom left)
|
|
ax_bar = fig.add_subplot(gs[1:, :2])
|
|
top_15 = langs_sorted[:15]
|
|
y_pos = np.arange(15)
|
|
bar_colors = [colors_by_cat.get(categories[i], '#666') for i in range(15)]
|
|
ax_bar.barh(y_pos, [l['duration_seconds'] for l in top_15], color=bar_colors, edgecolor='#fff')
|
|
ax_bar.set_yticks(y_pos)
|
|
ax_bar.set_yticklabels([l['language'] for l in top_15], fontsize=9)
|
|
ax_bar.invert_yaxis()
|
|
ax_bar.set_xlabel('Duration (seconds)')
|
|
ax_bar.set_title('Slowest 15 Languages', fontsize=12, color='#fff', fontweight='bold')
|
|
|
|
# Pie chart (bottom right)
|
|
ax_pie = fig.add_subplot(gs[1, 2])
|
|
ax_pie.pie([compiled_count, interpreted_count],
|
|
labels=['Compiled', 'Interpreted'],
|
|
autopct='%1.0f%%',
|
|
colors=['#e94560', '#0f3460'],
|
|
textprops={'color': '#eee'},
|
|
wedgeprops={'edgecolor': '#fff'})
|
|
ax_pie.set_title('Language Types', fontsize=12, color='#fff', fontweight='bold')
|
|
|
|
# Stats text (bottom right corner)
|
|
ax_stats = fig.add_subplot(gs[2, 2])
|
|
ax_stats.axis('off')
|
|
stats = f"""
|
|
Fastest: {summary['fastest_language']} ({summary['min_duration_seconds']}s)
|
|
Slowest: {summary['slowest_language']} ({summary['max_duration_seconds']}s)
|
|
Spread: {summary['max_duration_seconds'] - summary['min_duration_seconds']}s
|
|
|
|
Generated: {data['generated_at'][:10]}
|
|
Pipeline: #{data['child_pipeline_id']}
|
|
"""
|
|
ax_stats.text(0.1, 0.9, stats, fontsize=11, color='#aaa', va='top',
|
|
transform=ax_stats.transAxes, family='monospace')
|
|
|
|
plt.savefig('/tmp/artifacts/chart-dashboard.png', dpi=150, facecolor='#1a1a2e')
|
|
print('✓ chart-dashboard.png')
|
|
|
|
print(f'\n✅ Generated 6 charts in /tmp/artifacts/')
|
|
print('Files:')
|
|
for f in os.listdir('/tmp/artifacts'):
|
|
print(f' - {f}')
|