TIOVX User Guide
J722S Linux Performance Report
SoC J722S
Test EVM J722S TI EVM
HOST ARM A53 running Linux at default frequency
Target DSP C7X 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) 2.564000 2.444000
2 AccumulateSquare 640x480 (307200) 2.697000 2.593000
3 AccumulateWeighted 640x480 (307200) 1.170000 1.069000
4 Addition U8+U8=U8 640x480 (307200) 1.219000 1.112000
5 Addition U8+U8=S16 640x480 (307200) 2.714000 2.608000
6 Addition U8+S16=S16 640x480 (307200) 3.268000 3.161000
7 Addition S16+S16=S16 640x480 (307200) 4.168000 4.063000
8 Subtraction U8-U8=U8 640x480 (307200) 1.220000 1.115000
9 Subtraction U8-U8=S16 640x480 (307200) 2.459000 2.360000
10 Subtraction S16-U8=S16 640x480 (307200) 4.441000 4.337000
11 Subtraction S16-S16=S16 640x480 (307200) 4.171000 4.068000
12 Not Operation U8 640x480 (307200) 1.160000 1.057000
13 And Operation U8 640x480 (307200) 2.146000 2.043000
14 OR Operation U8 640x480 (307200) 2.143000 2.047000
15 XOR Operation U8 640x480 (307200) 2.146000 2.044000
16 Abs Diff U8 640x480 (307200) 1.330000 1.229000
17 Abs Diff S16 640x480 (307200) 2.450000 2.350000
18 Box 3x3 640x480 (307200) 1.763000 1.660000
19 Dilate 3x3 640x480 (307200) 1.604000 1.498000
20 Erode 3x3 640x480 (307200) 1.590000 1.485000
21 Fast Corners No NMS 640x480 (307200) 18.364000 18.240000
22 Canny 640x480 (307200) 19.289000 19.169000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 4.594000 4.481000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 2.298000 2.194000
25 Channel Combine YUYV 640x480 (307200) 1.939000 1.836000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 1.851000 1.742000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 3.347000 3.244000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 4.474000 4.370000
29 Color Convert RGB=>RGBX 640x480 (307200) 3.463000 3.357000
30 Color Convert RGB=>NV12 640x480 (307200) 3.720000 3.610000
31 Color Convert RGB=>IYUV 640x480 (307200) 4.006000 3.898000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.984000 3.880000
33 Color Convert RGBX=>RGB 640x480 (307200) 4.921000 4.813000
34 Color Convert RGBX=>NV12 640x480 (307200) 4.667000 4.558000
35 Color Convert RGBX=>IYUV 640x480 (307200) 4.966000 4.862000
36 Color Convert RGBX=>YUV4 640x480 (307200) 5.013000 4.905000
37 Color Convert NVXX=>RGB 640x480 (307200) 3.895000 3.784000
38 Color Convert NVXX=>RGBX 640x480 (307200) 3.803000 3.697000
39 Color Convert NVXX=>IYUV 640x480 (307200) 1.836000 1.727000
40 Color Convert NVXX=>YUV4 640x480 (307200) 1.864000 1.750000
41 Color Convert YUVX=>RGB 640x480 (307200) 3.458000 3.349000
42 Color Convert YUVX=>RGBX 640x480 (307200) 3.196000 3.091000
43 Color Convert YUVX=>NV12 640x480 (307200) 3.404000 3.295000
44 Color Convert YUVX=>IYUV 640x480 (307200) 4.442000 4.332000
45 Color Convert IYUV=>RGB 640x480 (307200) 3.655000 3.543000
46 Color Convert IYUV=>RGBX 640x480 (307200) 3.041000 2.932000
47 Color Convert IYUV=>NV12 640x480 (307200) 1.871000 1.759000
48 Color Convert IYUV=>YUV4 640x480 (307200) 2.166000 2.063000
49 Convert Depth U8 to S16 640x480 (307200) 1.467000 1.362000
50 Convert Depth S16 to U8 640x480 (307200) 2.475000 2.371000
51 Convolution 3x3, O: U8 640x480 (307200) 2.571000 2.468000
52 Convolution 3x3, O: S16 640x480 (307200) 2.964000 2.857000
53 Convolution 9x3, O: U8 640x480 (307200) 46.517000 46.372000
54 Convolution 9x3, O: S16 640x480 (307200) 47.009000 46.869000
55 Convolution 3x9, O: U8 640x480 (307200) 67.781000 67.607000
56 Convolution 3x9, O: S16 640x480 (307200) 67.726000 67.612000
57 Convolution 5x5, O: U8 640x480 (307200) 4.208000 4.099000
58 Convolution 5x5, O: S16 640x480 (307200) 4.407000 4.301000
59 Convolution 7x7, O: U8 640x480 (307200) 4.841000 4.726000
60 Convolution 7x7, O: S16 640x480 (307200) 4.851000 4.742000
61 Convolution 9x9, O: U8 640x480 (307200) 117.303000 117.172000
62 Convolution 9x9, O: S16 640x480 (307200) 118.881000 118.741000
63 Equalize Histogram 640x480 (307200) 3.179000 3.075000
64 Gaussian 3x3 640x480 (307200) 1.636000 1.534000
65 Gaussian Pyramid 640x480 (307200) 14.257000 14.140000
66 Half Scale Gaussian Pyramid 640x480 (307200) 1.387000 1.285000
67 Harris Corners 640x480 (307200) 30.426000 30.302000
68 Histogram U8 640x480 (307200) 1.737000 1.622000
69 Integral Image 640x480 (307200) 6.379000 6.273000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 9.123000 9.009000
71 Laplacian Reconstruct U8 640x480 (307200) 11.089000 10.984000
72 LookUpTable S16 640x480 (307200) 2.820000 2.717000
73 LookUpTable U8 640x480 (307200) 1.610000 1.501000
74 Magnitude S16 640x480 (307200) 4.958000 4.853000
75 Mean/Standard Deviation U8 640x480 (307200) 0.880000 0.779000
76 Median 3x3 640x480 (307200) 1.652000 1.552000
77 Min Max Loc U8 640x480 (307200) 1.670000 1.563000
78 Min Max Loc S16 640x480 (307200) 3.180000 3.073000
79 Multiply U8 x U8 = U8 640x480 (307200) 1.380000 1.280000
80 Multiply U8 x U8 = S16 640x480 (307200) 1.663000 1.560000
81 Multiply U8 x S16 = S16 640x480 (307200) 3.700000 3.594000
82 Multiply S16 x S16 = S16 640x480 (307200) 4.009000 3.905000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 3.066000 2.959000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 3.062000 2.955000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 62.394000 62.272000
86 Optical Flow 5x5 640x480 (307200) 5.973000 5.868000
87 Phase 640x480 (307200) 5.920000 5.815000
88 Remap Bilinear Interpolation *640x480 (307200) 13.647000 13.543000
89 Remap Nearest Neighbor *640x480 (307200) 9.329000 9.221000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.820000 1.713000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 6.369000 6.264000
92 Sobel 3x3 640x480 (307200) 4.483000 4.366000
93 Threshold Binary 640x480 (307200) 1.257000 1.154000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.430000 0.319000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 1.254000 1.150000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.615000 0.511000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.878000 1.775000

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