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) 2.622000 2.531000
2 AccumulateSquare 640x480 (307200) 2.656000 2.563000
3 AccumulateWeighted 640x480 (307200) 1.745000 1.654000
4 Addition U8+U8=U8 640x480 (307200) 2.103000 2.011000
5 Addition U8+U8=S16 640x480 (307200) 3.058000 2.963000
6 Addition U8+S16=S16 640x480 (307200) 4.197000 4.101000
7 Addition S16+S16=S16 640x480 (307200) 3.810000 3.711000
8 Subtraction U8-U8=U8 640x480 (307200) 2.573000 2.482000
9 Subtraction U8-U8=S16 640x480 (307200) 3.097000 3.002000
10 Subtraction S16-U8=S16 640x480 (307200) 3.847000 3.755000
11 Subtraction S16-S16=S16 640x480 (307200) 4.897000 4.788000
12 Not Operation U8 640x480 (307200) 1.562000 1.472000
13 And Operation U8 640x480 (307200) 2.070000 1.977000
14 OR Operation U8 640x480 (307200) 2.037000 1.943000
15 XOR Operation U8 640x480 (307200) 2.038000 1.946000
16 Abs Diff U8 640x480 (307200) 2.533000 2.444000
17 Abs Diff S16 640x480 (307200) 4.827000 4.734000
18 Box 3x3 640x480 (307200) 1.711000 1.623000
19 Dilate 3x3 640x480 (307200) 1.754000 1.659000
20 Erode 3x3 640x480 (307200) 1.755000 1.665000
21 Fast Corners No NMS 640x480 (307200) 13.641000 13.541000
22 Canny 640x480 (307200) 15.176000 15.072000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 4.852000 4.755000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 6.309000 6.213000
25 Channel Combine YUYV 640x480 (307200) 3.053000 2.960000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 2.502000 2.408000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 2.727000 2.636000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 3.918000 3.828000
29 Color Convert RGB=>RGBX 640x480 (307200) 5.281000 5.187000
30 Color Convert RGB=>NV12 640x480 (307200) 3.842000 3.748000
31 Color Convert RGB=>IYUV 640x480 (307200) 3.968000 3.873000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.958000 3.866000
33 Color Convert RGBX=>RGB 640x480 (307200) 5.045000 4.950000
34 Color Convert RGBX=>NV12 640x480 (307200) 4.665000 4.570000
35 Color Convert RGBX=>IYUV 640x480 (307200) 4.682000 4.586000
36 Color Convert RGBX=>YUV4 640x480 (307200) 4.733000 4.639000
37 Color Convert NVXX=>RGB 640x480 (307200) 3.745000 3.649000
38 Color Convert NVXX=>RGBX 640x480 (307200) 4.744000 4.649000
39 Color Convert NVXX=>IYUV 640x480 (307200) 2.404000 2.310000
40 Color Convert NVXX=>YUV4 640x480 (307200) 3.069000 2.974000
41 Color Convert YUVX=>RGB 640x480 (307200) 4.469000 4.376000
42 Color Convert YUVX=>RGBX 640x480 (307200) 5.282000 5.188000
43 Color Convert YUVX=>NV12 640x480 (307200) 3.576000 3.481000
44 Color Convert YUVX=>IYUV 640x480 (307200) 3.336000 3.240000
45 Color Convert IYUV=>RGB 640x480 (307200) 3.988000 3.894000
46 Color Convert IYUV=>RGBX 640x480 (307200) 4.791000 4.698000
47 Color Convert IYUV=>NV12 640x480 (307200) 2.494000 2.405000
48 Color Convert IYUV=>YUV4 640x480 (307200) 3.367000 3.273000
49 Convert Depth U8 to S16 640x480 (307200) 2.563000 2.467000
50 Convert Depth S16 to U8 640x480 (307200) 2.573000 2.480000
51 Convolution 3x3, O: U8 640x480 (307200) 1.974000 1.881000
52 Convolution 3x3, O: S16 640x480 (307200) 2.730000 2.635000
53 Convolution 9x3, O: U8 640x480 (307200) 18.973000 18.876000
54 Convolution 9x3, O: S16 640x480 (307200) 18.477000 18.379000
55 Convolution 3x9, O: U8 640x480 (307200) 24.579000 24.480000
56 Convolution 3x9, O: S16 640x480 (307200) 23.380000 23.264000
57 Convolution 5x5, O: U8 640x480 (307200) 2.339000 2.246000
58 Convolution 5x5, O: S16 640x480 (307200) 2.837000 2.744000
59 Convolution 7x7, O: U8 640x480 (307200) 3.198000 3.100000
60 Convolution 7x7, O: S16 640x480 (307200) 3.416000 3.326000
61 Convolution 9x9, O: U8 640x480 (307200) 37.661000 37.563000
62 Convolution 9x9, O: S16 640x480 (307200) 36.482000 36.383000
63 Equalize Histogram 640x480 (307200) 2.600000 2.507000
64 Gaussian 3x3 640x480 (307200) 1.829000 1.738000
65 Gaussian Pyramid 640x480 (307200) 10.359000 10.252000
66 Half Scale Gaussian Pyramid 640x480 (307200) 1.197000 1.106000
67 Harris Corners 640x480 (307200) 29.580000 29.473000
68 Histogram U8 640x480 (307200) 1.119000 1.027000
69 Integral Image 640x480 (307200) 4.571000 4.476000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 7.544000 7.431000
71 Laplacian Reconstruct U8 640x480 (307200) 10.407000 10.300000
72 LookUpTable S16 640x480 (307200) 3.319000 3.224000
73 LookUpTable U8 640x480 (307200) 1.840000 1.749000
74 Magnitude S16 640x480 (307200) 5.981000 5.887000
75 Mean/Standard Deviation U8 640x480 (307200) 0.929000 0.836000
76 Median 3x3 640x480 (307200) 1.878000 1.780000
77 Min Max Loc U8 640x480 (307200) 1.590000 1.493000
78 Min Max Loc S16 640x480 (307200) 2.963000 2.865000
79 Multiply U8 x U8 = U8 640x480 (307200) 2.093000 1.996000
80 Multiply U8 x U8 = S16 640x480 (307200) 2.015000 1.920000
81 Multiply U8 x S16 = S16 640x480 (307200) 4.379000 4.285000
82 Multiply S16 x S16 = S16 640x480 (307200) 3.844000 3.747000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.672000 2.578000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.649000 2.556000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 22.658000 22.559000
86 Optical Flow 5x5 640x480 (307200) 5.304000 5.202000
87 Phase 640x480 (307200) 5.155000 5.061000
88 Remap Bilinear Interpolation *640x480 (307200) 8.861000 8.764000
89 Remap Nearest Neighbor *640x480 (307200) 7.697000 7.597000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.212000 1.120000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 2.585000 2.482000
92 Sobel 3x3 640x480 (307200) 4.939000 4.844000
93 Threshold Binary 640x480 (307200) 1.736000 1.642000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.538000 0.446000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 0.825000 0.733000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.710000 0.616000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.181000 1.086000

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