TIOVX User Guide
J721E Linux Performance Report
SoC J721E
Test EVM J721E TI EVM
HOST ARM A72 running Linux at default frequency
Target DSP C66x 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) 2.694000 2.601000
2 AccumulateSquare 640x480 (307200) 2.732000 2.652000
3 AccumulateWeighted 640x480 (307200) 1.921000 1.847000
4 Addition U8+U8=U8 640x480 (307200) 1.973000 1.898000
5 Addition U8+U8=S16 640x480 (307200) 3.204000 3.125000
6 Addition U8+S16=S16 640x480 (307200) 4.000000 3.922000
7 Addition S16+S16=S16 640x480 (307200) 3.815000 3.735000
8 Subtraction U8-U8=U8 640x480 (307200) 2.536000 2.458000
9 Subtraction U8-U8=S16 640x480 (307200) 3.201000 3.121000
10 Subtraction S16-U8=S16 640x480 (307200) 4.012000 3.935000
11 Subtraction S16-S16=S16 640x480 (307200) 4.907000 4.828000
12 Not Operation U8 640x480 (307200) 1.663000 1.588000
13 And Operation U8 640x480 (307200) 1.967000 1.893000
14 OR Operation U8 640x480 (307200) 1.968000 1.894000
15 XOR Operation U8 640x480 (307200) 1.963000 1.887000
16 Abs Diff U8 640x480 (307200) 2.534000 2.455000
17 Abs Diff S16 640x480 (307200) 4.892000 4.812000
18 Box 3x3 640x480 (307200) 1.791000 1.716000
19 Dilate 3x3 640x480 (307200) 1.647000 1.559000
20 Erode 3x3 640x480 (307200) 1.634000 1.560000
21 Fast Corners No NMS 640x480 (307200) 13.404000 13.318000
22 Canny 640x480 (307200) 14.802000 14.715000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 5.508000 5.424000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 7.391000 7.307000
25 Channel Combine YUYV 640x480 (307200) 2.976000 2.892000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 2.376000 2.296000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 2.616000 2.542000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 3.882000 3.804000
29 Color Convert RGB=>RGBX 640x480 (307200) 5.147000 5.067000
30 Color Convert RGB=>NV12 640x480 (307200) 3.788000 3.709000
31 Color Convert RGB=>IYUV 640x480 (307200) 3.852000 3.771000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.776000 3.696000
33 Color Convert RGBX=>RGB 640x480 (307200) 4.940000 4.861000
34 Color Convert RGBX=>NV12 640x480 (307200) 4.556000 4.472000
35 Color Convert RGBX=>IYUV 640x480 (307200) 4.632000 4.548000
36 Color Convert RGBX=>YUV4 640x480 (307200) 4.642000 4.533000
37 Color Convert NVXX=>RGB 640x480 (307200) 3.633000 3.553000
38 Color Convert NVXX=>RGBX 640x480 (307200) 4.531000 4.450000
39 Color Convert NVXX=>IYUV 640x480 (307200) 2.336000 2.263000
40 Color Convert NVXX=>YUV4 640x480 (307200) 2.864000 2.786000
41 Color Convert YUVX=>RGB 640x480 (307200) 4.289000 4.209000
42 Color Convert YUVX=>RGBX 640x480 (307200) 5.258000 5.178000
43 Color Convert YUVX=>NV12 640x480 (307200) 3.492000 3.411000
44 Color Convert YUVX=>IYUV 640x480 (307200) 3.254000 3.172000
45 Color Convert IYUV=>RGB 640x480 (307200) 3.745000 3.667000
46 Color Convert IYUV=>RGBX 640x480 (307200) 4.555000 4.475000
47 Color Convert IYUV=>NV12 640x480 (307200) 2.385000 2.307000
48 Color Convert IYUV=>YUV4 640x480 (307200) 3.115000 3.034000
49 Convert Depth U8 to S16 640x480 (307200) 2.404000 2.327000
50 Convert Depth S16 to U8 640x480 (307200) 2.439000 2.363000
51 Convolution 3x3, O: U8 640x480 (307200) 1.852000 1.776000
52 Convolution 3x3, O: S16 640x480 (307200) 2.567000 2.491000
53 Convolution 9x3, O: U8 640x480 (307200) 18.802000 18.722000
54 Convolution 9x3, O: S16 640x480 (307200) 18.283000 18.203000
55 Convolution 3x9, O: U8 640x480 (307200) 24.479000 24.376000
56 Convolution 3x9, O: S16 640x480 (307200) 23.204000 23.125000
57 Convolution 5x5, O: U8 640x480 (307200) 2.188000 2.110000
58 Convolution 5x5, O: S16 640x480 (307200) 2.661000 2.584000
59 Convolution 7x7, O: U8 640x480 (307200) 3.006000 2.931000
60 Convolution 7x7, O: S16 640x480 (307200) 3.214000 3.138000
61 Convolution 9x9, O: U8 640x480 (307200) 37.493000 37.411000
62 Convolution 9x9, O: S16 640x480 (307200) 36.269000 36.191000
63 Equalize Histogram 640x480 (307200) 2.603000 2.527000
64 Gaussian 3x3 640x480 (307200) 1.693000 1.618000
65 Gaussian Pyramid 640x480 (307200) 10.465000 10.375000
66 Half Scale Gaussian Pyramid 640x480 (307200) 1.080000 1.000000
67 Harris Corners 640x480 (307200) 29.281000 29.189000
68 Histogram U8 640x480 (307200) 1.034000 0.956000
69 Integral Image 640x480 (307200) 4.095000 4.014000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 7.595000 7.500000
71 Laplacian Reconstruct U8 640x480 (307200) 10.285000 10.197000
72 LookUpTable U8 640x480 (307200) 3.422000 3.342000
73 LookUpTable U8 640x480 (307200) 1.699000 1.623000
74 Magnitude S16 640x480 (307200) 5.656000 5.576000
75 Mean/Standard Deviation U8 640x480 (307200) 0.852000 0.776000
76 Median 3x3 640x480 (307200) 1.749000 1.674000
77 Min Max Loc U8 640x480 (307200) 1.552000 1.471000
78 Min Max Loc S16 640x480 (307200) 2.999000 2.916000
79 Multiply U8 x U8 = U8 640x480 (307200) 2.617000 2.539000
80 Multiply U8 x U8 = S16 640x480 (307200) 2.097000 2.017000
81 Multiply U8 x S16 = S16 640x480 (307200) 4.128000 4.047000
82 Multiply S16 x S16 = S16 640x480 (307200) 3.820000 3.739000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.499000 2.418000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.496000 2.413000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 22.470000 22.388000
86 Optical Flow 5x5 640x480 (307200) 5.311000 5.225000
87 Phase 640x480 (307200) 5.136000 5.056000
88 Remap Bilinear Interpolation *640x480 (307200) 8.785000 8.701000
89 Remap Nearest Neighbor *640x480 (307200) 7.574000 7.493000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.076000 0.998000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 2.453000 2.377000
92 Sobel 3x3 640x480 (307200) 4.742000 4.663000
93 Threshold Binary 640x480 (307200) 1.646000 1.565000
94 Warp Affine Nearest Neighbor *640x480 (307200) 0.426000 0.352000
95 Warp Affine Bilinear Interpolation *640x480 (307200) 0.842000 0.764000
96 Warp Pespective Nearest Neighbor *640x480 (307200) 0.624000 0.546000
97 Warp Pespective Bilinear Interpolation *640x480 (307200) 1.072000 0.994000

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