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) 2.555000 2.453000
2 AccumulateSquare 640x480 (307200) 2.676000 2.575000
3 AccumulateWeighted 640x480 (307200) 1.175000 1.071000
4 Addition U8+U8=U8 640x480 (307200) 1.212000 1.110000
5 Addition U8+U8=S16 640x480 (307200) 2.702000 2.604000
6 Addition U8+S16=S16 640x480 (307200) 3.242000 3.142000
7 Addition S16+S16=S16 640x480 (307200) 4.174000 4.073000
8 Subtraction U8-U8=U8 640x480 (307200) 1.209000 1.106000
9 Subtraction U8-U8=S16 640x480 (307200) 2.476000 2.373000
10 Subtraction S16-U8=S16 640x480 (307200) 3.541000 3.437000
11 Subtraction S16-S16=S16 640x480 (307200) 4.182000 4.079000
12 Not Operation U8 640x480 (307200) 1.158000 1.055000
13 And Operation U8 640x480 (307200) 2.144000 2.043000
14 OR Operation U8 640x480 (307200) 2.139000 2.037000
15 XOR Operation U8 640x480 (307200) 2.144000 2.042000
16 Abs Diff U8 640x480 (307200) 1.316000 1.214000
17 Abs Diff S16 640x480 (307200) 2.469000 2.366000
18 Box 3x3 640x480 (307200) 1.760000 1.657000
19 Dilate 3x3 640x480 (307200) 1.623000 1.516000
20 Erode 3x3 640x480 (307200) 1.628000 1.525000
21 Fast Corners No NMS 640x480 (307200) 18.334000 18.224000
22 Canny 640x480 (307200) 19.297000 19.159000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 4.319000 4.214000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 1.918000 1.810000
25 Channel Combine YUYV 640x480 (307200) 1.945000 1.841000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 1.861000 1.754000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 3.361000 3.256000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 4.491000 4.387000
29 Color Convert RGB=>RGBX 640x480 (307200) 3.460000 3.341000
30 Color Convert RGB=>NV12 640x480 (307200) 3.757000 3.648000
31 Color Convert RGB=>IYUV 640x480 (307200) 4.047000 3.940000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.993000 3.883000
33 Color Convert RGBX=>RGB 640x480 (307200) 4.954000 4.845000
34 Color Convert RGBX=>NV12 640x480 (307200) 4.668000 4.560000
35 Color Convert RGBX=>IYUV 640x480 (307200) 4.973000 4.868000
36 Color Convert RGBX=>YUV4 640x480 (307200) 5.004000 4.900000
37 Color Convert NVXX=>RGB 640x480 (307200) 3.897000 3.788000
38 Color Convert NVXX=>RGBX 640x480 (307200) 3.806000 3.701000
39 Color Convert NVXX=>IYUV 640x480 (307200) 1.813000 1.707000
40 Color Convert NVXX=>YUV4 640x480 (307200) 1.841000 1.738000
41 Color Convert YUVX=>RGB 640x480 (307200) 3.477000 3.370000
42 Color Convert YUVX=>RGBX 640x480 (307200) 3.195000 3.089000
43 Color Convert YUVX=>NV12 640x480 (307200) 3.393000 3.289000
44 Color Convert YUVX=>IYUV 640x480 (307200) 4.431000 4.325000
45 Color Convert IYUV=>RGB 640x480 (307200) 3.660000 3.551000
46 Color Convert IYUV=>RGBX 640x480 (307200) 3.035000 2.931000
47 Color Convert IYUV=>NV12 640x480 (307200) 1.840000 1.735000
48 Color Convert IYUV=>YUV4 640x480 (307200) 2.168000 2.061000
49 Convert Depth U8 to S16 640x480 (307200) 1.454000 1.351000
50 Convert Depth S16 to U8 640x480 (307200) 2.496000 2.393000
51 Convolution 3x3, O: U8 640x480 (307200) 2.695000 2.588000
52 Convolution 3x3, O: S16 640x480 (307200) 2.984000 2.880000
53 Convolution 9x3, O: U8 640x480 (307200) 46.486000 46.371000
54 Convolution 9x3, O: S16 640x480 (307200) 46.976000 46.862000
55 Convolution 3x9, O: U8 640x480 (307200) 67.796000 67.640000
56 Convolution 3x9, O: S16 640x480 (307200) 67.151000 67.036000
57 Convolution 5x5, O: U8 640x480 (307200) 4.313000 4.207000
58 Convolution 5x5, O: S16 640x480 (307200) 4.437000 4.331000
59 Convolution 7x7, O: U8 640x480 (307200) 4.840000 4.735000
60 Convolution 7x7, O: S16 640x480 (307200) 4.851000 4.743000
61 Convolution 9x9, O: U8 640x480 (307200) 117.297000 117.162000
62 Convolution 9x9, O: S16 640x480 (307200) 118.885000 118.665000
63 Equalize Histogram 640x480 (307200) 3.204000 3.103000
64 Gaussian 3x3 640x480 (307200) 1.664000 1.558000
65 Gaussian Pyramid 640x480 (307200) 14.059000 13.947000
66 Half Scale Gaussian Pyramid 640x480 (307200) 1.395000 1.285000
67 Harris Corners 640x480 (307200) 30.461000 30.343000
68 Histogram U8 640x480 (307200) 1.721000 1.619000
69 Integral Image 640x480 (307200) 6.412000 6.285000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 9.118000 9.010000
71 Laplacian Reconstruct U8 640x480 (307200) 11.165000 11.067000
72 LookUpTable U8 640x480 (307200) 2.810000 2.710000
73 LookUpTable U8 640x480 (307200) 1.637000 1.525000
74 Magnitude S16 640x480 (307200) 4.966000 4.863000
75 Mean/Standard Deviation U8 640x480 (307200) 0.880000 0.780000
76 Median 3x3 640x480 (307200) 1.694000 1.594000
77 Min Max Loc U8 640x480 (307200) 1.661000 1.562000
78 Min Max Loc S16 640x480 (307200) 3.175000 3.072000
79 Multiply U8 x U8 = U8 640x480 (307200) 1.349000 1.250000
80 Multiply U8 x U8 = S16 640x480 (307200) 1.668000 1.566000
81 Multiply U8 x S16 = S16 640x480 (307200) 3.640000 3.541000
82 Multiply S16 x S16 = S16 640x480 (307200) 4.008000 3.907000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 3.135000 3.032000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 3.129000 3.027000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 62.383000 62.273000
86 Optical Flow 5x5 640x480 (307200) 5.928000 5.807000
87 Phase 640x480 (307200) 5.914000 5.813000
88 Remap Bilinear Interpolation *640x480 (307200) 13.658000 13.554000
89 Remap Nearest Neighbor *640x480 (307200) 9.306000 9.195000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.813000 1.714000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 6.372000 6.263000
92 Sobel 3x3 640x480 (307200) 4.438000 4.336000
93 Threshold Binary 640x480 (307200) 1.314000 1.212000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.419000 0.319000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 1.232000 1.133000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.622000 0.520000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.856000 1.757000

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