Skip to content

Latest commit

 

History

History
190 lines (153 loc) · 11.3 KB

lvgl_benchmark.md

File metadata and controls

190 lines (153 loc) · 11.3 KB
Raw

LVGL Benchmark

1 Overview

This is a benchmark report for comparing performance differences between LVGL baseline and LVGL plus arm-2d acceleration (taking advantage of Helium technology).

NOTE: Due to LVGL's own rendering schema, the result cannot be interpreted as Helium acceleration vs C implementation, it is not a fair comparison. For example, in the scene Image chorma keyed, the report shows almost 10x performance (987%), this is because in the baseline LVGL repeats its decoding-transform iterations for many times untill all lines are processed, and in arm-2d acceleration, the process is only performed once.

The result truly shows the advantage of using arm-2d to accelerate LVGL when Helium technology is available but doesn't not reveal the right performance ratio between a C verison and a Helium accelerated version for a given 2D image processing algorithm.

1.1 Test Environment

1.1.1 Hardware

  • MPS3
  • Processor: Cortex-M55 (r1p0)
  • System Frequency: 60MHz
  • LCD: 320*240 RGB565 8080 interface

1.1.2 Software

1.1.3 Configuration

  • LVGL (configured in lv_conf_cmsis.h)

    • set LV_COLOR_DEPTH to 16
    • set LV_MEM_SIZE to (64U * 1024U)
    • set LV_DISP_DEF_REFR_PERIOD to 2
    • RTE Configuration

    RTE

    • Source code
    static void on_benchmark_finished(void)
{
    disp_enable();
}

int main(void)
{
    
    lv_init();
    lv_port_disp_init();
    lv_port_indev_init();

    __LL_LCD_PRINT(24, 0, "Running LVGL Benchmark...");
    __LL_LCD_PRINT(25, 0, "Please stand by...");
    __LL_LCD_PRINT(28, 0, "NOTE: You will NOT see anything until the end.");

    disp_disable();
    
    lv_demo_benchmark_set_finished_cb(&on_benchmark_finished);
    lv_demo_benchmark();
    
    while(1) {
        lv_timer_handler();
    }
}

  • Arm-2D (configured in arm_2d_cfg.h)

    • set __ARM_2D_HAS_ASYNC__to 0
    • set __ARM_2D_HAS_ANTI_ALIAS_TRANSFORM__ to 1
    • set __ARM_2D_CFG_SUPPORT_COLOUR_CHANNEL_ACCESS__ to 1

2 Result

Table 2-1 LVGL Benchmark (baseline vs arm-2d acceleration)

No. Scene LVGL-baseline LVGL+Arm-2D Ratio
Weighted FPS 279 430 154.1%
Opa. speed 94% 65% 69.1%
0 Rectangle 204 231 113.2%
1 Rectangle + opa 100 141 141.0%
2 Rectangle rounded 131 139 106.1%
3 Rectangle rounded + opa 74 110 148.6%
4 Circle 69 74 107.2%
5 Circle + opa 44 73 165.9%
6 Border 181 207 114.4%
7 Border + opa 163 188 115.3%
8 Border rounded 121 133 109.9%
9 Border rounded + opa 109 127 116.5%
10 Circle border 59 61 103.4%
11 Circle border + opa 50 58 116.0%
12 Border top 142 145 102.1%
13 Border top + opa 135 143 105.9%
14 Border left 142 159 112.0%
15 Border left + opa 138 152 110.1%
16 Border top + left 124 131 105.6%
17 Border top + left + opa 116 125 107.8%
18 Border left + right 113 121 107.1%
19 Border left + right + opa 105 114 108.6%
20 Border top + bottom 120 121 100.8%
21 Border top + bottom + opa 114 121 106.1%
22 Shadow small 38 44 115.8%
23 Shadow small + opa 35 43 122.9%
24 Shadow small offset 40 47 117.5%
25 Shadow small offset + opa 33 43 130.3%
26 Shadow large 19 24 126.3%
27 Shadow large + opa 16 24 150.0%
28 Shadow large offset 20 25 125.0%
29 Shadow large offset + opa 17 25 147.1%
30 Image RGB 814 3918 481.3%
31 Image RGB + opa 424 1079 254.5%
32 Image ARGB 393 415 105.6%
33 Image ARGB + opa 310 405 130.6%
34 Image chorma keyed 428 4225 987.1%
35 Image chorma keyed + opa 312 1023 327.9%
36 Image indexed 231 235 101.7%
37 Image indexed + opa 199 228 114.6%
38 Image alpha only 218 216 99.1%
39 Image alpha only + opa 189 213 112.7%
40 Image RGB recolor 311 864 277.8%
41 Image RGB recolor + opa 247 563 227.9%
42 Image ARGB recolor 251 369 147.0%
43 Image ARGB recolor + opa 221 357 161.5%
44 Image chorma keyed recolor 262 856 326.7%
45 Image chorma keyed recolor + opa 218 532 244.0%
46 Image indexed recolor 182 220 120.9%
47 Image indexed recolor + opa 165 214 129.7%
48 Image RGB rotate 101 99 98.0%
49 Image RGB rotate + opa 75 98 130.7%
50 Image RGB rotate anti aliased 43 106 246.5%
51 Image RGB rotate anti aliased + opa 37 100 270.3%
52 Image ARGB rotate 88 91 103.4%
53 Image ARGB rotate + opa 77 90 116.9%
54 Image ARGB rotate anti aliased 37 37 100.0%
55 Image ARGB rotate anti aliased + opa 35 37 105.7%
56 Image RGB zoom 175 175 100.0%
57 Image RGB zoom + opa 134 176 131.3%
58 Image RGB zoom anti aliased 75 172 229.3%
59 Image RGB zoom anti aliased + opa 66 163 247.0%
60 Image ARGB zoom 154 162 105.2%
61 Image ARGB zoom + opa 137 163 119.0%
62 Image ARGB zoom anti aliased 65 66 101.5%
63 Image ARGB zoom anti aliased + opa 62 65 104.8%
64 Text small 15 16 106.7%
65 Text small + opa 15 17 113.3%
66 Text medium 15 16 106.7%
67 Text medium + opa 14 16 114.3%
68 Text large 15 16 106.7%
69 Text large + opa 14 16 114.3%
70 Text small compressed 0 0 0.0%
71 Text small compressed + opa 19 19 100.0%
72 Text medium compressed 19 18 94.7%
73 Text medium compressed + opa 19 19 100.0%
74 Text large compressed 22 23 104.5%
75 Text large compressed + opa 22 23 104.5%
76 Line 59 64 108.5%
77 Line + opa 57 66 115.8%
78 Arc think 88 90 102.3%
79 Arc think + opa 86 88 102.3%
80 Arc thick 90 92 102.2%
81 Arc thick + opa 85 92 108.2%
82 Substr. rectangle 242 251 103.7%
83 Substr. rectangle + opa 291 317 108.9%
84 Substr. border 291 319 109.6%
85 Substr. border + opa 286 311 108.7%
86 Substr. shadow 328 345 105.2%
87 Substr. shadow + opa 322 349 108.4%
88 Substr. image 5380 5282 98.2%
89 Substr. image + opa 5962 5247 88.0%
90 Substr. line 333 341 102.4%
91 Substr. line + opa 344 381 110.8%
92 Substr. arc 81 81 100.0%
93 Substr. arc + opa 80 80 100.0%
94 Substr. text 57 57 100.0%
95 Substr. text + opa 57 58 101.8%

Figure 2-1 Performance Ratio (Arm-2D acceleration over baseline)

Figure2-1