zenresize 0.2.0

High-quality image resampling with 31 filters, streaming API, and SIMD acceleration
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208

#!/usr/bin/env python3
"""Plot zenresize performance sweep: throughput vs input size, by ratio and feature mode."""

import csv
import sys
from collections import defaultdict
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import matplotlib.ticker as ticker
import numpy as np

def load_csv(path):
    rows = []
    with open(path) as f:
        reader = csv.DictReader(f)
        for row in reader:
            row['in_size'] = int(row['in_size'])
            row['in_pixels'] = int(row['in_pixels'])
            row['out_pixels'] = int(row['out_pixels'])
            row['mean_ms'] = float(row['mean_ms'])
            row['stddev_ms'] = float(row['stddev_ms'])
            row['in_mpps'] = float(row['in_mpps'])
            row['out_mpps'] = float(row['out_mpps'])
            rows.append(row)
    return rows

def main():
    safe_data = load_csv('/tmp/sweep_safe.csv')
    unsafe_data = load_csv('/tmp/sweep_unsafe.csv')

    import os
    outdir = os.environ.get('SWEEP_OUTPUT_DIR', '/mnt/v/output/zenresize/sweep')
    os.makedirs(outdir, exist_ok=True)

    # --- Chart 1: Throughput (input MP/s) vs input size, one line per ratio ---
    # Safe mode only, downscale ratios
    fig, ax = plt.subplots(figsize=(14, 8))

    ratio_order = ['12.5%', '25%', '33%', '50%', '67%', '75%', '100%']
    colors = plt.cm.viridis(np.linspace(0.1, 0.9, len(ratio_order)))

    for i, ratio in enumerate(ratio_order):
        pts = sorted([r for r in safe_data if r['ratio'] == ratio], key=lambda r: r['in_size'])
        if not pts:
            continue
        sizes = [p['in_size'] for p in pts]
        mpps = [p['in_mpps'] for p in pts]
        ax.plot(sizes, mpps, 'o-', color=colors[i], label=f'→{ratio}', linewidth=2, markersize=5)

    ax.set_xlabel('Input size (square, pixels per side)', fontsize=12)
    ax.set_ylabel('Throughput (input MP/s)', fontsize=12)
    ax.set_title('zenresize sRGB throughput vs input size (safe mode, downscale)', fontsize=14)
    ax.legend(title='Output ratio', fontsize=10)
    ax.grid(True, alpha=0.3)
    ax.set_xscale('log', base=2)
    ax.xaxis.set_major_formatter(ticker.FuncFormatter(lambda x, _: f'{int(x)}'))
    fig.tight_layout()
    fig.savefig(f'{outdir}/throughput_vs_size_safe.png', dpi=150)
    print(f'Saved {outdir}/throughput_vs_size_safe.png')
    plt.close()

    # --- Chart 2: Safe vs Unsafe throughput comparison at 50% downscale ---
    fig, ax = plt.subplots(figsize=(14, 8))

    for label, data, color, marker in [
        ('safe', safe_data, '#1f77b4', 'o'),
        ('unsafe_kernels', unsafe_data, '#ff7f0e', 's'),
    ]:
        pts = sorted([r for r in data if r['ratio'] == '50%'], key=lambda r: r['in_size'])
        sizes = [p['in_size'] for p in pts]
        mpps = [p['in_mpps'] for p in pts]
        ax.plot(sizes, mpps, f'{marker}-', color=color, label=label, linewidth=2, markersize=6)

    ax.set_xlabel('Input size (square, pixels per side)', fontsize=12)
    ax.set_ylabel('Throughput (input MP/s)', fontsize=12)
    ax.set_title('zenresize sRGB: safe vs unsafe_kernels (50% downscale)', fontsize=14)
    ax.legend(fontsize=12)
    ax.grid(True, alpha=0.3)
    ax.set_xscale('log', base=2)
    ax.xaxis.set_major_formatter(ticker.FuncFormatter(lambda x, _: f'{int(x)}'))
    fig.tight_layout()
    fig.savefig(f'{outdir}/safe_vs_unsafe_50pct.png', dpi=150)
    print(f'Saved {outdir}/safe_vs_unsafe_50pct.png')
    plt.close()

    # --- Chart 3: Throughput vs ratio at fixed sizes ---
    fig, axes = plt.subplots(2, 2, figsize=(16, 12))
    target_sizes = [1024, 2048, 3840, 7680]

    ratio_to_float = {
        '12.5%': 0.125, '25%': 0.25, '33%': 0.333, '50%': 0.5,
        '67%': 0.667, '75%': 0.75, '100%': 1.0, '150%': 1.5,
        '200%': 2.0, '300%': 3.0,
    }

    for ax, size in zip(axes.flat, target_sizes):
        for label, data, color, marker in [
            ('safe', safe_data, '#1f77b4', 'o'),
            ('unsafe_kernels', unsafe_data, '#ff7f0e', 's'),
        ]:
            pts = sorted(
                [r for r in data if r['in_size'] == size],
                key=lambda r: ratio_to_float.get(r['ratio'], 0)
            )
            if not pts:
                continue
            ratios = [ratio_to_float[p['ratio']] for p in pts]
            mpps = [p['in_mpps'] for p in pts]
            ax.plot(ratios, mpps, f'{marker}-', color=color, label=label, linewidth=2, markersize=6)

        ax.set_xlabel('Scale ratio', fontsize=10)
        ax.set_ylabel('Throughput (input MP/s)', fontsize=10)
        ax.set_title(f'{size}x{size}', fontsize=12)
        ax.legend(fontsize=9)
        ax.grid(True, alpha=0.3)
        ax.axvline(x=1.0, color='gray', linestyle='--', alpha=0.5)

    fig.suptitle('zenresize sRGB throughput vs scale ratio', fontsize=14)
    fig.tight_layout()
    fig.savefig(f'{outdir}/throughput_vs_ratio.png', dpi=150)
    print(f'Saved {outdir}/throughput_vs_ratio.png')
    plt.close()

    # --- Chart 4: Overhead heatmap (safe time / unsafe time) ---
    fig, ax = plt.subplots(figsize=(14, 8))

    # Build lookup for unsafe times
    unsafe_lookup = {}
    for r in unsafe_data:
        unsafe_lookup[(r['in_size'], r['ratio'])] = r['mean_ms']

    all_sizes = sorted(set(r['in_size'] for r in safe_data))
    all_ratios_str = ['12.5%', '25%', '33%', '50%', '67%', '75%', '100%', '150%', '200%', '300%']

    overhead_matrix = []
    valid_sizes = []
    valid_ratios = []

    for ratio in all_ratios_str:
        row = []
        for size in all_sizes:
            safe_ms = None
            unsafe_ms = None
            for r in safe_data:
                if r['in_size'] == size and r['ratio'] == ratio:
                    safe_ms = r['mean_ms']
                    break
            unsafe_ms = unsafe_lookup.get((size, ratio))
            if safe_ms is not None and unsafe_ms is not None and unsafe_ms > 0:
                row.append((safe_ms / unsafe_ms - 1.0) * 100.0)  # percent overhead
            else:
                row.append(float('nan'))
        overhead_matrix.append(row)

    overhead_arr = np.array(overhead_matrix)

    im = ax.imshow(overhead_arr, aspect='auto', cmap='RdYlGn_r', vmin=-5, vmax=25,
                   interpolation='nearest')
    ax.set_xticks(range(len(all_sizes)))
    ax.set_xticklabels([str(s) for s in all_sizes], rotation=45, ha='right', fontsize=8)
    ax.set_yticks(range(len(all_ratios_str)))
    ax.set_yticklabels(all_ratios_str, fontsize=10)
    ax.set_xlabel('Input size (square)', fontsize=12)
    ax.set_ylabel('Scale ratio', fontsize=12)
    ax.set_title('Bounds-checking overhead: (safe / unsafe_kernels − 1) × 100%', fontsize=14)

    # Annotate cells
    for i in range(len(all_ratios_str)):
        for j in range(len(all_sizes)):
            val = overhead_arr[i, j]
            if not np.isnan(val):
                color = 'white' if abs(val) > 15 else 'black'
                ax.text(j, i, f'{val:.0f}%', ha='center', va='center', fontsize=7, color=color)

    fig.colorbar(im, ax=ax, label='Overhead %')
    fig.tight_layout()
    fig.savefig(f'{outdir}/overhead_heatmap.png', dpi=150)
    print(f'Saved {outdir}/overhead_heatmap.png')
    plt.close()

    # --- Chart 5: Time vs input megapixels (log-log), all ratios ---
    fig, ax = plt.subplots(figsize=(14, 8))

    for i, ratio in enumerate(ratio_order):
        pts = sorted([r for r in safe_data if r['ratio'] == ratio], key=lambda r: r['in_pixels'])
        if not pts:
            continue
        mpx = [p['in_pixels'] / 1e6 for p in pts]
        ms = [p['mean_ms'] for p in pts]
        ax.plot(mpx, ms, 'o-', color=colors[i], label=f'→{ratio}', linewidth=2, markersize=5)

    ax.set_xlabel('Input megapixels', fontsize=12)
    ax.set_ylabel('Time (ms)', fontsize=12)
    ax.set_title('zenresize sRGB time vs input size (safe mode)', fontsize=14)
    ax.legend(title='Output ratio', fontsize=10)
    ax.grid(True, alpha=0.3)
    ax.set_xscale('log')
    ax.set_yscale('log')
    fig.tight_layout()
    fig.savefig(f'{outdir}/time_vs_mpx_loglog.png', dpi=150)
    print(f'Saved {outdir}/time_vs_mpx_loglog.png')
    plt.close()

    print('\nDone. All charts saved to', outdir)

if __name__ == '__main__':
    main()