TIOVX User Guide
J721S2 Linux Performance Report
SoC J721S2
Test EVM J721S2 TI EVM
HOST ARM A72 running Linux at default frequency
Target Standalone C7x DSP running Free RTOS at default frequency, L1/L2 Cache Enabled (Cache-only based performance.)



TIOVX Kernel Performance
Index Kernel Variant Frame Size (Pixels) Graph Performance (msec) Node Performance (msec)
1 Accumulate 640x480 (307200) 1.285000 1.163000
2 AccumulateSquare 640x480 (307200) 1.696000 1.587000
3 AccumulateWeighted 640x480 (307200) 1.236000 1.143000
4 Addition U8+U8=U8 640x480 (307200) 1.062000 0.961000
5 Addition U8+U8=S16 640x480 (307200) 1.146000 1.038000
6 Addition U8+S16=S16 640x480 (307200) 1.294000 1.178000
7 Addition S16+S16=S16 640x480 (307200) 2.341000 2.215000
8 Subtraction U8-U8=U8 640x480 (307200) 1.113000 1.004000
9 Subtraction U8-U8=S16 640x480 (307200) 1.105000 0.992000
10 Subtraction S16-U8=S16 640x480 (307200) 1.589000 1.474000
11 Subtraction S16-S16=S16 640x480 (307200) 2.319000 2.193000
12 Not Operation U8 640x480 (307200) 0.592000 0.495000
13 And Operation U8 640x480 (307200) 1.055000 0.954000
14 OR Operation U8 640x480 (307200) 1.032000 0.931000
15 XOR Operation U8 640x480 (307200) 0.984000 0.897000
16 Abs Diff U8 640x480 (307200) 1.069000 0.979000
17 Abs Diff S16 640x480 (307200) 2.421000 2.302000
18 Box 3x3 640x480 (307200) 1.285000 1.189000
19 Dilate 3x3 640x480 (307200) 0.757000 0.654000
20 Erode 3x3 640x480 (307200) 0.746000 0.636000
21 Fast Corners No NMS 640x480 (307200) 13.699000 13.558000
22 Canny 640x480 (307200) 8.491000 8.343000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 2.315000 2.173000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 3.004000 2.848000
25 Channel Combine YUYV 640x480 (307200) 1.370000 1.250000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 0.950000 0.836000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 1.781000 1.656000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 2.185000 2.056000
29 Color Convert RGB=>RGBX 640x480 (307200) 3.366000 3.229000
30 Color Convert RGB=>NV12 640x480 (307200) 2.948000 2.822000
31 Color Convert RGB=>IYUV 640x480 (307200) 2.892000 2.764000
32 Color Convert RGB=>YUV4 640x480 (307200) 2.984000 2.847000
33 Color Convert RGBX=>RGB 640x480 (307200) 3.354000 3.213000
34 Color Convert RGBX=>NV12 640x480 (307200) 3.400000 3.264000
35 Color Convert RGBX=>IYUV 640x480 (307200) 3.616000 3.472000
36 Color Convert RGBX=>YUV4 640x480 (307200) 3.410000 3.277000
37 Color Convert NVXX=>RGB 640x480 (307200) 2.727000 2.595000
38 Color Convert NVXX=>RGBX 640x480 (307200) 3.335000 3.173000
39 Color Convert NVXX=>IYUV 640x480 (307200) 1.020000 0.894000
40 Color Convert NVXX=>YUV4 640x480 (307200) 1.049000 0.920000
41 Color Convert YUVX=>RGB 640x480 (307200) 3.039000 2.904000
42 Color Convert YUVX=>RGBX 640x480 (307200) 2.640000 2.502000
43 Color Convert YUVX=>NV12 640x480 (307200) 1.796000 1.670000
44 Color Convert YUVX=>IYUV 640x480 (307200) 1.782000 1.658000
45 Color Convert IYUV=>RGB 640x480 (307200) 2.692000 2.561000
46 Color Convert IYUV=>RGBX 640x480 (307200) 2.760000 2.630000
47 Color Convert IYUV=>NV12 640x480 (307200) 1.017000 0.894000
48 Color Convert IYUV=>YUV4 640x480 (307200) 0.945000 0.811000
49 Convert Depth U8 to S16 640x480 (307200) 0.713000 0.599000
50 Convert Depth S16 to U8 640x480 (307200) 1.138000 1.029000
51 Convolution 3x3, O: U8 640x480 (307200) 1.392000 1.280000
52 Convolution 3x3, O: S16 640x480 (307200) 1.273000 1.147000
53 Convolution 9x3, O: U8 640x480 (307200) 47.485000 47.379000
54 Convolution 9x3, O: S16 640x480 (307200) 48.233000 48.125000
55 Convolution 3x9, O: U8 640x480 (307200) 62.615000 62.515000
56 Convolution 3x9, O: S16 640x480 (307200) 63.582000 63.467000
57 Convolution 5x5, O: U8 640x480 (307200) 2.471000 2.366000
58 Convolution 5x5, O: S16 640x480 (307200) 2.456000 2.344000
59 Convolution 7x7, O: U8 640x480 (307200) 9.441000 9.334000
60 Convolution 7x7, O: S16 640x480 (307200) 4.061000 3.944000
61 Convolution 9x9, O: U8 640x480 (307200) 121.916000 121.813000
62 Convolution 9x9, O: S16 640x480 (307200) 122.880000 122.774000
63 Equalize Histogram 640x480 (307200) 1.902000 1.801000
64 Gaussian 3x3 640x480 (307200) 0.889000 0.787000
65 Gaussian Pyramid 640x480 (307200) 7.781000 7.640000
66 Half Scale Gaussian Pyramid 640x480 (307200) 0.700000 0.603000
67 Harris Corners 640x480 (307200) 17.425000 17.258000
68 Histogram U8 640x480 (307200) 1.332000 1.226000
69 Integral Image 640x480 (307200) 1.312000 1.176000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 3.652000 3.498000
71 Laplacian Reconstruct U8 640x480 (307200) 4.755000 4.602000
72 LookUpTable U8 640x480 (307200) 2.086000 1.955000
73 LookUpTable U8 640x480 (307200) 0.971000 0.854000
74 Magnitude S16 640x480 (307200) 5.103000 4.956000
75 Mean/Standard Deviation U8 640x480 (307200) 0.661000 0.563000
76 Median 3x3 640x480 (307200) 1.322000 1.212000
77 Min Max Loc U8 640x480 (307200) 1.017000 0.919000
78 Min Max Loc S16 640x480 (307200) 2.091000 1.971000
79 Multiply U8 x U8 = U8 640x480 (307200) 1.320000 1.209000
80 Multiply U8 x U8 = S16 640x480 (307200) 0.941000 0.837000
81 Multiply U8 x S16 = S16 640x480 (307200) 1.986000 1.867000
82 Multiply S16 x S16 = S16 640x480 (307200) 3.510000 3.386000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.961000 2.849000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.937000 2.835000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 60.372000 60.269000
86 Optical Flow 5x5 640x480 (307200) 3.795000 3.655000
87 Phase 640x480 (307200) 5.871000 5.732000
88 Remap Bilinear Interpolation *640x480 (307200) 9.705000 9.550000
89 Remap Nearest Neighbor *640x480 (307200) 8.010000 7.865000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.345000 1.242000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 3.684000 3.596000
92 Sobel 3x3 640x480 (307200) 1.624000 1.478000
93 Threshold Binary 640x480 (307200) 0.683000 0.579000
94 Warp Affine Nearest Neighbor *640x480 (307200) 0.364000 0.277000
95 Warp Affine Bilinear Interpolation *640x480 (307200) 1.004000 0.912000
96 Warp Pespective Nearest Neighbor *640x480 (307200) 0.599000 0.510000
97 Warp Pespective Bilinear Interpolation *640x480 (307200) 1.745000 1.653000

Footnote 1: All optional parameters for relevant kernels are used

Footnote 2: All image sizes are for input parameters unless noted with an asterisk (*) in which case these are the output parameter sizes