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) 2.642000 2.561000
2 AccumulateSquare 640x480 (307200) 2.673000 2.593000
3 AccumulateWeighted 640x480 (307200) 1.738000 1.661000
4 Addition U8+U8=U8 640x480 (307200) 2.075000 1.997000
5 Addition U8+U8=S16 640x480 (307200) 3.039000 2.959000
6 Addition U8+S16=S16 640x480 (307200) 4.135000 4.054000
7 Addition S16+S16=S16 640x480 (307200) 3.795000 3.716000
8 Subtraction U8-U8=U8 640x480 (307200) 2.554000 2.471000
9 Subtraction U8-U8=S16 640x480 (307200) 3.077000 2.997000
10 Subtraction S16-U8=S16 640x480 (307200) 4.017000 3.934000
11 Subtraction S16-S16=S16 640x480 (307200) 4.881000 4.794000
12 Not Operation U8 640x480 (307200) 1.553000 1.471000
13 And Operation U8 640x480 (307200) 2.032000 1.955000
14 OR Operation U8 640x480 (307200) 2.021000 1.941000
15 XOR Operation U8 640x480 (307200) 2.025000 1.944000
16 Abs Diff U8 640x480 (307200) 2.527000 2.449000
17 Abs Diff S16 640x480 (307200) 4.800000 4.719000
18 Box 3x3 640x480 (307200) 1.699000 1.621000
19 Dilate 3x3 640x480 (307200) 1.671000 1.590000
20 Erode 3x3 640x480 (307200) 1.678000 1.598000
21 Fast Corners No NMS 640x480 (307200) 13.549000 13.462000
22 Canny 640x480 (307200) 15.094000 15.002000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 4.799000 4.716000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 6.177000 6.096000
25 Channel Combine YUYV 640x480 (307200) 3.022000 2.939000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 2.414000 2.332000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 2.694000 2.613000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 3.926000 3.843000
29 Color Convert RGB=>RGBX 640x480 (307200) 5.291000 5.212000
30 Color Convert RGB=>NV12 640x480 (307200) 3.827000 3.743000
31 Color Convert RGB=>IYUV 640x480 (307200) 3.965000 3.881000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.968000 3.887000
33 Color Convert RGBX=>RGB 640x480 (307200) 5.043000 4.960000
34 Color Convert RGBX=>NV12 640x480 (307200) 4.661000 4.577000
35 Color Convert RGBX=>IYUV 640x480 (307200) 4.694000 4.611000
36 Color Convert RGBX=>YUV4 640x480 (307200) 4.737000 4.655000
37 Color Convert NVXX=>RGB 640x480 (307200) 3.697000 3.617000
38 Color Convert NVXX=>RGBX 640x480 (307200) 4.695000 4.615000
39 Color Convert NVXX=>IYUV 640x480 (307200) 2.362000 2.282000
40 Color Convert NVXX=>YUV4 640x480 (307200) 3.039000 2.958000
41 Color Convert YUVX=>RGB 640x480 (307200) 4.456000 4.375000
42 Color Convert YUVX=>RGBX 640x480 (307200) 5.224000 5.140000
43 Color Convert YUVX=>NV12 640x480 (307200) 3.578000 3.501000
44 Color Convert YUVX=>IYUV 640x480 (307200) 3.339000 3.255000
45 Color Convert IYUV=>RGB 640x480 (307200) 3.917000 3.836000
46 Color Convert IYUV=>RGBX 640x480 (307200) 4.706000 4.626000
47 Color Convert IYUV=>NV12 640x480 (307200) 2.430000 2.349000
48 Color Convert IYUV=>YUV4 640x480 (307200) 3.300000 3.219000
49 Convert Depth U8 to S16 640x480 (307200) 2.484000 2.398000
50 Convert Depth S16 to U8 640x480 (307200) 2.549000 2.463000
51 Convolution 3x3, O: U8 640x480 (307200) 1.894000 1.811000
52 Convolution 3x3, O: S16 640x480 (307200) 2.605000 2.524000
53 Convolution 9x3, O: U8 640x480 (307200) 18.892000 18.809000
54 Convolution 9x3, O: S16 640x480 (307200) 18.389000 18.305000
55 Convolution 3x9, O: U8 640x480 (307200) 24.493000 24.406000
56 Convolution 3x9, O: S16 640x480 (307200) 23.327000 23.235000
57 Convolution 5x5, O: U8 640x480 (307200) 2.240000 2.160000
58 Convolution 5x5, O: S16 640x480 (307200) 2.719000 2.641000
59 Convolution 7x7, O: U8 640x480 (307200) 3.101000 3.020000
60 Convolution 7x7, O: S16 640x480 (307200) 3.301000 3.222000
61 Convolution 9x9, O: U8 640x480 (307200) 37.585000 37.502000
62 Convolution 9x9, O: S16 640x480 (307200) 36.403000 36.312000
63 Equalize Histogram 640x480 (307200) 2.636000 2.554000
64 Gaussian 3x3 640x480 (307200) 1.747000 1.664000
65 Gaussian Pyramid 640x480 (307200) 10.300000 10.196000
66 Half Scale Gaussian Pyramid 640x480 (307200) 1.129000 1.028000
67 Harris Corners 640x480 (307200) 29.583000 29.487000
68 Histogram U8 640x480 (307200) 1.081000 0.997000
69 Integral Image 640x480 (307200) 4.495000 4.409000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 7.540000 7.429000
71 Laplacian Reconstruct U8 640x480 (307200) 10.440000 10.344000
72 LookUpTable U8 640x480 (307200) 3.273000 3.190000
73 LookUpTable U8 640x480 (307200) 1.744000 1.663000
74 Magnitude S16 640x480 (307200) 5.737000 5.655000
75 Mean/Standard Deviation U8 640x480 (307200) 0.906000 0.817000
76 Median 3x3 640x480 (307200) 1.820000 1.733000
77 Min Max Loc U8 640x480 (307200) 1.609000 1.526000
78 Min Max Loc S16 640x480 (307200) 3.015000 2.929000
79 Multiply U8 x U8 = U8 640x480 (307200) 2.094000 2.008000
80 Multiply U8 x U8 = S16 640x480 (307200) 2.049000 1.962000
81 Multiply U8 x S16 = S16 640x480 (307200) 4.394000 4.309000
82 Multiply S16 x S16 = S16 640x480 (307200) 3.858000 3.774000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.564000 2.480000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.546000 2.464000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 22.586000 22.496000
86 Optical Flow 5x5 640x480 (307200) 5.341000 5.249000
87 Phase 640x480 (307200) 5.032000 4.949000
88 Remap Bilinear Interpolation *640x480 (307200) 8.950000 8.864000
89 Remap Nearest Neighbor *640x480 (307200) 7.739000 7.649000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.160000 1.080000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 2.512000 2.432000
92 Sobel 3x3 640x480 (307200) 4.836000 4.754000
93 Threshold Binary 640x480 (307200) 1.666000 1.582000
94 Warp Affine Nearest Neighbor *640x480 (307200) 0.471000 0.389000
95 Warp Affine Bilinear Interpolation *640x480 (307200) 0.802000 0.720000
96 Warp Pespective Nearest Neighbor *640x480 (307200) 0.665000 0.584000
97 Warp Pespective Bilinear Interpolation *640x480 (307200) 1.131000 1.048000

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