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) 1.238000 1.115000
2 AccumulateSquare 640x480 (307200) 1.768000 1.644000
3 AccumulateWeighted 640x480 (307200) 0.927000 0.820000
4 Addition U8+U8=U8 640x480 (307200) 0.776000 0.647000
5 Addition U8+U8=S16 640x480 (307200) 0.891000 0.750000
6 Addition U8+S16=S16 640x480 (307200) 1.279000 1.136000
7 Addition S16+S16=S16 640x480 (307200) 1.333000 1.182000
8 Subtraction U8-U8=U8 640x480 (307200) 0.771000 0.639000
9 Subtraction U8-U8=S16 640x480 (307200) 0.856000 0.715000
10 Subtraction S16-U8=S16 640x480 (307200) 1.767000 1.621000
11 Subtraction S16-S16=S16 640x480 (307200) 1.384000 1.195000
12 Not Operation U8 640x480 (307200) 0.595000 0.483000
13 And Operation U8 640x480 (307200) 0.738000 0.612000
14 OR Operation U8 640x480 (307200) 0.727000 0.605000
15 XOR Operation U8 640x480 (307200) 0.728000 0.601000
16 Abs Diff U8 640x480 (307200) 0.771000 0.656000
17 Abs Diff S16 640x480 (307200) 1.535000 1.386000
18 Box 3x3 640x480 (307200) 1.375000 1.258000
19 Dilate 3x3 640x480 (307200) 0.764000 0.638000
20 Erode 3x3 640x480 (307200) 0.757000 0.635000
21 Fast Corners No NMS 640x480 (307200) 13.678000 13.493000
22 Canny 640x480 (307200) 7.880000 7.702000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 1.529000 1.361000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 1.747000 1.561000
25 Channel Combine YUYV 640x480 (307200) 1.124000 0.981000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 1.011000 0.874000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 1.786000 1.651000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 2.326000 2.181000
29 Color Convert RGB=>RGBX 640x480 (307200) 3.570000 3.398000
30 Color Convert RGB=>NV12 640x480 (307200) 3.185000 3.028000
31 Color Convert RGB=>IYUV 640x480 (307200) 3.110000 2.947000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.126000 2.976000
33 Color Convert RGBX=>RGB 640x480 (307200) 3.637000 3.475000
34 Color Convert RGBX=>NV12 640x480 (307200) 3.661000 3.507000
35 Color Convert RGBX=>IYUV 640x480 (307200) 3.615000 3.453000
36 Color Convert RGBX=>YUV4 640x480 (307200) 3.630000 3.461000
37 Color Convert NVXX=>RGB 640x480 (307200) 2.863000 2.703000
38 Color Convert NVXX=>RGBX 640x480 (307200) 3.588000 3.424000
39 Color Convert NVXX=>IYUV 640x480 (307200) 1.027000 0.884000
40 Color Convert NVXX=>YUV4 640x480 (307200) 1.170000 1.008000
41 Color Convert YUVX=>RGB 640x480 (307200) 3.221000 3.062000
42 Color Convert YUVX=>RGBX 640x480 (307200) 2.816000 2.651000
43 Color Convert YUVX=>NV12 640x480 (307200) 1.889000 1.740000
44 Color Convert YUVX=>IYUV 640x480 (307200) 1.870000 1.721000
45 Color Convert IYUV=>RGB 640x480 (307200) 2.919000 2.765000
46 Color Convert IYUV=>RGBX 640x480 (307200) 3.015000 2.852000
47 Color Convert IYUV=>NV12 640x480 (307200) 1.080000 0.937000
48 Color Convert IYUV=>YUV4 640x480 (307200) 1.043000 0.898000
49 Convert Depth U8 to S16 640x480 (307200) 0.721000 0.587000
50 Convert Depth S16 to U8 640x480 (307200) 1.252000 1.119000
51 Convolution 3x3, O: U8 640x480 (307200) 1.314000 1.192000
52 Convolution 3x3, O: S16 640x480 (307200) 1.283000 1.151000
53 Convolution 9x3, O: U8 640x480 (307200) 47.400000 47.271000
54 Convolution 9x3, O: S16 640x480 (307200) 48.273000 48.128000
55 Convolution 3x9, O: U8 640x480 (307200) 63.058000 62.933000
56 Convolution 3x9, O: S16 640x480 (307200) 63.773000 63.638000
57 Convolution 5x5, O: U8 640x480 (307200) 2.420000 2.297000
58 Convolution 5x5, O: S16 640x480 (307200) 2.463000 2.330000
59 Convolution 7x7, O: U8 640x480 (307200) 3.809000 3.693000
60 Convolution 7x7, O: S16 640x480 (307200) 4.214000 4.061000
61 Convolution 9x9, O: U8 640x480 (307200) 121.928000 121.808000
62 Convolution 9x9, O: S16 640x480 (307200) 122.894000 122.740000
63 Equalize Histogram 640x480 (307200) 1.985000 1.834000
64 Gaussian 3x3 640x480 (307200) 0.924000 0.804000
65 Gaussian Pyramid 640x480 (307200) 7.730000 7.538000
66 Half Scale Gaussian Pyramid 640x480 (307200) 0.713000 0.601000
67 Harris Corners 640x480 (307200) 17.672000 17.482000
68 Histogram U8 640x480 (307200) 1.280000 1.179000
69 Integral Image 640x480 (307200) 1.321000 1.165000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 3.719000 3.547000
71 Laplacian Reconstruct U8 640x480 (307200) 4.539000 4.359000
72 LookUpTable S16 640x480 (307200) 1.968000 1.819000
73 LookUpTable U8 640x480 (307200) 0.902000 0.787000
74 Magnitude S16 640x480 (307200) 3.915000 3.749000
75 Mean/Standard Deviation U8 640x480 (307200) 0.615000 0.512000
76 Median 3x3 640x480 (307200) 1.350000 1.233000
77 Min Max Loc U8 640x480 (307200) 1.011000 0.907000
78 Min Max Loc S16 640x480 (307200) 1.991000 1.872000
79 Multiply U8 x U8 = U8 640x480 (307200) 0.870000 0.753000
80 Multiply U8 x U8 = S16 640x480 (307200) 0.958000 0.816000
81 Multiply U8 x S16 = S16 640x480 (307200) 2.046000 1.899000
82 Multiply S16 x S16 = S16 640x480 (307200) 2.535000 2.377000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.955000 2.781000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.910000 2.780000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 60.352000 60.215000
86 Optical Flow 5x5 640x480 (307200) 3.806000 3.647000
87 Phase 640x480 (307200) 4.808000 4.663000
88 Remap Bilinear Interpolation *640x480 (307200) 9.880000 9.696000
89 Remap Nearest Neighbor *640x480 (307200) 8.537000 8.358000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.362000 1.239000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 5.246000 5.136000
92 Sobel 3x3 640x480 (307200) 1.618000 1.464000
93 Threshold Binary 640x480 (307200) 0.697000 0.577000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.362000 0.270000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 1.144000 1.044000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.620000 0.517000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.782000 1.681000

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