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) 2.651000 2.552000
2 AccumulateSquare 640x480 (307200) 2.676000 2.579000
3 AccumulateWeighted 640x480 (307200) 1.743000 1.650000
4 Addition U8+U8=U8 640x480 (307200) 2.109000 2.015000
5 Addition U8+U8=S16 640x480 (307200) 3.137000 3.040000
6 Addition U8+S16=S16 640x480 (307200) 4.252000 4.155000
7 Addition S16+S16=S16 640x480 (307200) 3.605000 3.506000
8 Subtraction U8-U8=U8 640x480 (307200) 2.600000 2.502000
9 Subtraction U8-U8=S16 640x480 (307200) 3.103000 3.007000
10 Subtraction S16-U8=S16 640x480 (307200) 3.844000 3.742000
11 Subtraction S16-S16=S16 640x480 (307200) 4.568000 4.469000
12 Not Operation U8 640x480 (307200) 1.572000 1.479000
13 And Operation U8 640x480 (307200) 2.053000 1.960000
14 OR Operation U8 640x480 (307200) 2.031000 1.937000
15 XOR Operation U8 640x480 (307200) 2.044000 1.948000
16 Abs Diff U8 640x480 (307200) 2.567000 2.474000
17 Abs Diff S16 640x480 (307200) 4.494000 4.399000
18 Box 3x3 640x480 (307200) 1.717000 1.623000
19 Dilate 3x3 640x480 (307200) 1.770000 1.674000
20 Erode 3x3 640x480 (307200) 1.759000 1.667000
21 Fast Corners No NMS 640x480 (307200) 13.647000 13.549000
22 Canny 640x480 (307200) 15.186000 15.067000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 4.708000 4.610000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 6.196000 6.093000
25 Channel Combine YUYV 640x480 (307200) 3.012000 2.915000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 2.445000 2.348000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 2.679000 2.584000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 3.928000 3.829000
29 Color Convert RGB=>RGBX 640x480 (307200) 5.310000 5.207000
30 Color Convert RGB=>NV12 640x480 (307200) 3.787000 3.690000
31 Color Convert RGB=>IYUV 640x480 (307200) 3.978000 3.880000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.929000 3.827000
33 Color Convert RGBX=>RGB 640x480 (307200) 5.032000 4.932000
34 Color Convert RGBX=>NV12 640x480 (307200) 4.667000 4.566000
35 Color Convert RGBX=>IYUV 640x480 (307200) 4.779000 4.681000
36 Color Convert RGBX=>YUV4 640x480 (307200) 4.796000 4.699000
37 Color Convert NVXX=>RGB 640x480 (307200) 3.922000 3.824000
38 Color Convert NVXX=>RGBX 640x480 (307200) 4.723000 4.619000
39 Color Convert NVXX=>IYUV 640x480 (307200) 2.529000 2.433000
40 Color Convert NVXX=>YUV4 640x480 (307200) 3.004000 2.905000
41 Color Convert YUVX=>RGB 640x480 (307200) 4.491000 4.394000
42 Color Convert YUVX=>RGBX 640x480 (307200) 5.169000 5.070000
43 Color Convert YUVX=>NV12 640x480 (307200) 3.573000 3.473000
44 Color Convert YUVX=>IYUV 640x480 (307200) 3.418000 3.261000
45 Color Convert IYUV=>RGB 640x480 (307200) 3.801000 3.704000
46 Color Convert IYUV=>RGBX 640x480 (307200) 4.694000 4.593000
47 Color Convert IYUV=>NV12 640x480 (307200) 2.563000 2.466000
48 Color Convert IYUV=>YUV4 640x480 (307200) 3.218000 3.121000
49 Convert Depth U8 to S16 640x480 (307200) 2.603000 2.490000
50 Convert Depth S16 to U8 640x480 (307200) 2.529000 2.429000
51 Convolution 3x3, O: U8 640x480 (307200) 1.996000 1.899000
52 Convolution 3x3, O: S16 640x480 (307200) 2.733000 2.638000
53 Convolution 9x3, O: U8 640x480 (307200) 18.989000 18.886000
54 Convolution 9x3, O: S16 640x480 (307200) 18.477000 18.381000
55 Convolution 3x9, O: U8 640x480 (307200) 24.610000 24.509000
56 Convolution 3x9, O: S16 640x480 (307200) 23.379000 23.277000
57 Convolution 5x5, O: U8 640x480 (307200) 2.357000 2.261000
58 Convolution 5x5, O: S16 640x480 (307200) 2.855000 2.760000
59 Convolution 7x7, O: U8 640x480 (307200) 3.207000 3.109000
60 Convolution 7x7, O: S16 640x480 (307200) 3.437000 3.339000
61 Convolution 9x9, O: U8 640x480 (307200) 37.680000 37.576000
62 Convolution 9x9, O: S16 640x480 (307200) 36.509000 36.399000
63 Equalize Histogram 640x480 (307200) 2.602000 2.508000
64 Gaussian 3x3 640x480 (307200) 1.847000 1.751000
65 Gaussian Pyramid 640x480 (307200) 10.290000 10.178000
66 Half Scale Gaussian Pyramid 640x480 (307200) 1.207000 1.111000
67 Harris Corners 640x480 (307200) 29.589000 29.479000
68 Histogram U8 640x480 (307200) 1.130000 1.035000
69 Integral Image 640x480 (307200) 4.317000 4.223000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 7.783000 7.671000
71 Laplacian Reconstruct U8 640x480 (307200) 10.385000 10.275000
72 LookUpTable S16 640x480 (307200) 3.296000 3.201000
73 LookUpTable U8 640x480 (307200) 1.855000 1.760000
74 Magnitude S16 640x480 (307200) 5.554000 5.455000
75 Mean/Standard Deviation U8 640x480 (307200) 0.920000 0.826000
76 Median 3x3 640x480 (307200) 1.868000 1.773000
77 Min Max Loc U8 640x480 (307200) 1.577000 1.482000
78 Min Max Loc S16 640x480 (307200) 2.965000 2.866000
79 Multiply U8 x U8 = U8 640x480 (307200) 2.086000 1.991000
80 Multiply U8 x U8 = S16 640x480 (307200) 2.018000 1.922000
81 Multiply U8 x S16 = S16 640x480 (307200) 4.380000 4.280000
82 Multiply S16 x S16 = S16 640x480 (307200) 3.587000 3.490000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.632000 2.536000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.607000 2.511000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 22.668000 22.565000
86 Optical Flow 5x5 640x480 (307200) 5.296000 5.192000
87 Phase 640x480 (307200) 5.139000 5.043000
88 Remap Bilinear Interpolation *640x480 (307200) 8.936000 8.831000
89 Remap Nearest Neighbor *640x480 (307200) 7.726000 7.622000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.210000 1.112000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 2.566000 2.470000
92 Sobel 3x3 640x480 (307200) 4.962000 4.866000
93 Threshold Binary 640x480 (307200) 1.674000 1.578000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.529000 0.436000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 0.826000 0.730000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.709000 0.613000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.188000 1.094000

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