Datasheet

This chapter describes the performance measurements of the Edge AI Inference demos.

Performance data of the demos can be auto generated by running following command on target:

root@tda4vm-sk:/opt/edge_ai_apps/tests# ./gen_data_sheet.sh

The performence measurements includes the following

  1. FPS : Effective framerate at which the application runs

  2. Total time : Average time taken to process each frame, which includes pre-processing, inference and post-processing time

  3. Inference time : Average time taken to infer each frame

  4. CPU loading : Loading on different CPU cores present

  5. DDR BW : DDR read and write BW used

  6. HWA Loading : Loading on different Hardware accelerators present

Following are the latest performance numbers of the C++ demos:

Source : USB Camera

Capture Framerate : 30 fps Resolution : 720p format : JPEG

_images/edgeai_object_detection.png

Fig. 28 GStreamer based data-flow pipeline with USB camera input and display output

Model

FPS

Total time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ONR-CL-6158-mobileNetV2-1p4-qat

30.84

33.24

3.03

14.88

1523

539

2062

9.0

20.0

9.0

4.0

1.0

14.49

0

0

0

0

0

0

TFL-CL-0000-mobileNetV1-mlperf

30.63

33.15

1.06

23.9

1378

509

1887

5.0

22.0

9.0

4.0

1.0

15.27

0

0

0

0

0

0

TFL-OD-2020-ssdLite-mobDet-DSP-coco-320x320

30.82

33.23

5.00

18.93

1470

510

1980

15.0

28.0

9.0

4.0

1.0

14.56

0

0

0

0

0

0

Source : Video

Video Framerate : 30 fps Resolution : 720p Encoding : h264

_images/edgeai_video_source.png

Fig. 29 GStreamer based data-flow pipeline with video file input source and display output

Model

FPS

Total time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ONR-CL-6158-mobileNetV2-1p4-qat

30.59

33.09

3.76

38.1

1796

661

2457

11.0

20.0

9.0

6.0

1.0

17.20

0

0

0

0

0

0

TFL-CL-0000-mobileNetV1-mlperf

30.51

33.10

1.67

37.2

1643

629

2272

7.0

22.0

9.0

5.0

1.0

17.16

0

0

0

0

0

0

TFL-OD-2020-ssdLite-mobDet-DSP-coco-320x320

30.48

33.16

5.25

33.7

1739

626

2365

16.0

29.0

9.0

6.0

1.0

17.27

0

0

0

0

0

0

Source : CSI Camera (ov5640)

Capture Framerate : 30 fps Resolution : 720p format : YUYV

_images/edgeai_ov5640_camera_source.png

Fig. 30 GStreamer based data-flow pipeline for with CSI camera (OV5640) input and display output

Model

FPS

Total time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ONR-CL-6158-mobileNetV2-1p4-qat

29.60

34.11

3.02

12.53

1617

634

2251

9.0

45.0

9.0

5.0

1.0

17.57

0

0

0

0

0

0

TFL-CL-0000-mobileNetV1-mlperf

29.53

34.14

1.01

9.34

1469

604

2073

5.0

47.0

9.0

5.0

1.0

16.33

0

0

0

0

0

0

TFL-OD-2020-ssdLite-mobDet-DSP-coco-320x320

29.50

34.06

5.00

10.64

1545

597

2142

14.0

53.0

9.0

5.0

1.0

21.19

0

0

0

0

0

0

Source : CSI Camera with VISS (imx219)

Capture Framerate : 30 fps Resolution : 1080p format : SRGGB8

_images/edgeai_rpi_camera_source.png

Fig. 31 GStreamer based data-flow pipeline with IMX219 sensor, ISP and display

Model

FPS

Total time (ms)

Inference time (ms)

A72 Load (%)

DDR Read BW (MB/s)

DDR Write BW (MB/s)

DDR Total BW (MB/s)

C71 Load (%)

C66_1 Load (%)

C66_2 Load (%)

MCU2_0 Load (%)

MCU2_1 Load (%)

MSC_0 (%)

MSC_1 (%)

VISS (%)

NF (%)

LDC (%)

SDE (%)

DOF (%)

ONR-CL-6158-mobileNetV2-1p4-qat

30.72

33.20

3.01

14.32

1648

679

2327

9.0

16.0

9.0

8.0

1.0

24.42

0

11.25

0

0

0

0

TFL-CL-0000-mobileNetV1-mlperf

30.90

33.07

1.23

13.81

1492

650

2142

5.0

18.0

9.0

8.0

1.0

24.68

0

11.15

0

0

0

0

TFL-OD-2020-ssdLite-mobDet-DSP-coco-320x320

30.78

33.10

5.01

11.94

1566

643

2209

15.0

25.0

9.0

9.0

1.0

30.49

0

11.17

0

0

0

0