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) 2.505000 2.394000
2 AccumulateSquare 640x480 (307200) 2.570000 2.462000
3 AccumulateWeighted 640x480 (307200) 1.159000 1.050000
4 Addition U8+U8=U8 640x480 (307200) 1.179000 1.065000
5 Addition U8+U8=S16 640x480 (307200) 2.646000 2.529000
6 Addition U8+S16=S16 640x480 (307200) 3.138000 3.027000
7 Addition S16+S16=S16 640x480 (307200) 4.053000 3.941000
8 Subtraction U8-U8=U8 640x480 (307200) 1.176000 1.064000
9 Subtraction U8-U8=S16 640x480 (307200) 2.370000 2.237000
10 Subtraction S16-U8=S16 640x480 (307200) 3.448000 3.335000
11 Subtraction S16-S16=S16 640x480 (307200) 4.058000 3.940000
12 Not Operation U8 640x480 (307200) 1.125000 1.015000
13 And Operation U8 640x480 (307200) 2.102000 1.990000
14 OR Operation U8 640x480 (307200) 2.110000 1.989000
15 XOR Operation U8 640x480 (307200) 2.100000 1.988000
16 Abs Diff U8 640x480 (307200) 1.978000 1.867000
17 Abs Diff S16 640x480 (307200) 2.506000 2.399000
18 Box 3x3 640x480 (307200) 1.737000 1.624000
19 Dilate 3x3 640x480 (307200) 1.584000 1.471000
20 Erode 3x3 640x480 (307200) 1.600000 1.480000
21 Fast Corners No NMS 640x480 (307200) 17.806000 17.666000
22 Canny 640x480 (307200) 18.774000 18.651000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 1.979000 1.865000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 1.895000 1.779000
25 Channel Combine YUYV 640x480 (307200) 1.973000 1.855000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 1.718000 1.601000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 3.343000 3.222000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 4.185000 4.063000
29 Color Convert RGB=>RGBX 640x480 (307200) 3.374000 3.259000
30 Color Convert RGB=>NV12 640x480 (307200) 3.955000 3.836000
31 Color Convert RGB=>IYUV 640x480 (307200) 4.238000 4.116000
32 Color Convert RGB=>YUV4 640x480 (307200) 4.039000 3.925000
33 Color Convert RGBX=>RGB 640x480 (307200) 4.818000 4.702000
34 Color Convert RGBX=>NV12 640x480 (307200) 4.673000 4.552000
35 Color Convert RGBX=>IYUV 640x480 (307200) 4.947000 4.826000
36 Color Convert RGBX=>YUV4 640x480 (307200) 4.838000 4.721000
37 Color Convert NVXX=>RGB 640x480 (307200) 3.714000 3.592000
38 Color Convert NVXX=>RGBX 640x480 (307200) 3.649000 3.524000
39 Color Convert NVXX=>IYUV 640x480 (307200) 1.815000 1.700000
40 Color Convert NVXX=>YUV4 640x480 (307200) 1.817000 1.697000
41 Color Convert YUVX=>RGB 640x480 (307200) 3.887000 3.763000
42 Color Convert YUVX=>RGBX 640x480 (307200) 3.802000 3.684000
43 Color Convert YUVX=>NV12 640x480 (307200) 3.503000 3.384000
44 Color Convert YUVX=>IYUV 640x480 (307200) 3.514000 3.398000
45 Color Convert IYUV=>RGB 640x480 (307200) 3.758000 3.640000
46 Color Convert IYUV=>RGBX 640x480 (307200) 2.985000 2.869000
47 Color Convert IYUV=>NV12 640x480 (307200) 1.763000 1.650000
48 Color Convert IYUV=>YUV4 640x480 (307200) 1.900000 1.790000
49 Convert Depth U8 to S16 640x480 (307200) 1.442000 1.330000
50 Convert Depth S16 to U8 640x480 (307200) 2.444000 2.330000
51 Convolution 3x3, O: U8 640x480 (307200) 2.478000 2.355000
52 Convolution 3x3, O: S16 640x480 (307200) 2.830000 2.705000
53 Convolution 9x3, O: U8 640x480 (307200) 46.522000 46.392000
54 Convolution 9x3, O: S16 640x480 (307200) 46.946000 46.823000
55 Convolution 3x9, O: U8 640x480 (307200) 67.540000 67.413000
56 Convolution 3x9, O: S16 640x480 (307200) 68.707000 68.556000
57 Convolution 5x5, O: U8 640x480 (307200) 3.984000 3.857000
58 Convolution 5x5, O: S16 640x480 (307200) 4.179000 4.062000
59 Convolution 7x7, O: U8 640x480 (307200) 4.842000 4.731000
60 Convolution 7x7, O: S16 640x480 (307200) 4.856000 4.740000
61 Convolution 9x9, O: U8 640x480 (307200) 117.279000 117.136000
62 Convolution 9x9, O: S16 640x480 (307200) 118.811000 118.686000
63 Equalize Histogram 640x480 (307200) 3.130000 3.005000
64 Gaussian 3x3 640x480 (307200) 1.642000 1.524000
65 Gaussian Pyramid 640x480 (307200) 13.659000 13.536000
66 Half Scale Gaussian Pyramid 640x480 (307200) 1.317000 1.205000
67 Harris Corners 640x480 (307200) 29.704000 29.573000
68 Histogram U8 640x480 (307200) 1.744000 1.629000
69 Integral Image 640x480 (307200) 6.127000 5.998000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 8.334000 8.212000
71 Laplacian Reconstruct U8 640x480 (307200) 10.879000 10.760000
72 LookUpTable S16 640x480 (307200) 2.741000 2.618000
73 LookUpTable U8 640x480 (307200) 1.539000 1.422000
74 Magnitude S16 640x480 (307200) 5.064000 4.927000
75 Mean/Standard Deviation U8 640x480 (307200) 0.854000 0.745000
76 Median 3x3 640x480 (307200) 1.682000 1.567000
77 Min Max Loc U8 640x480 (307200) 1.598000 1.487000
78 Min Max Loc S16 640x480 (307200) 3.055000 2.935000
79 Multiply U8 x U8 = U8 640x480 (307200) 1.399000 1.281000
80 Multiply U8 x U8 = S16 640x480 (307200) 1.640000 1.525000
81 Multiply U8 x S16 = S16 640x480 (307200) 3.585000 3.460000
82 Multiply S16 x S16 = S16 640x480 (307200) 3.996000 3.881000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 3.090000 2.972000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 3.097000 2.974000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 62.284000 62.152000
86 Optical Flow 5x5 640x480 (307200) 5.608000 5.492000
87 Phase 640x480 (307200) 5.846000 5.724000
88 Remap Bilinear Interpolation *640x480 (307200) 13.317000 13.198000
89 Remap Nearest Neighbor *640x480 (307200) 9.124000 8.986000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.744000 1.632000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 6.182000 6.059000
92 Sobel 3x3 640x480 (307200) 4.313000 4.197000
93 Threshold Binary 640x480 (307200) 1.256000 1.142000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.442000 0.330000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 1.243000 1.124000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.628000 0.515000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.789000 1.674000

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