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) 2.658000 2.557000
2 AccumulateSquare 640x480 (307200) 2.680000 2.589000
3 AccumulateWeighted 640x480 (307200) 1.769000 1.661000
4 Addition U8+U8=U8 640x480 (307200) 2.090000 1.996000
5 Addition U8+U8=S16 640x480 (307200) 3.117000 3.022000
6 Addition U8+S16=S16 640x480 (307200) 4.240000 4.141000
7 Addition S16+S16=S16 640x480 (307200) 3.601000 3.503000
8 Subtraction U8-U8=U8 640x480 (307200) 2.600000 2.502000
9 Subtraction U8-U8=S16 640x480 (307200) 3.109000 3.012000
10 Subtraction S16-U8=S16 640x480 (307200) 3.851000 3.751000
11 Subtraction S16-S16=S16 640x480 (307200) 4.573000 4.471000
12 Not Operation U8 640x480 (307200) 1.583000 1.490000
13 And Operation U8 640x480 (307200) 2.049000 1.958000
14 OR Operation U8 640x480 (307200) 2.028000 1.940000
15 XOR Operation U8 640x480 (307200) 2.034000 1.940000
16 Abs Diff U8 640x480 (307200) 2.568000 2.477000
17 Abs Diff S16 640x480 (307200) 4.499000 4.408000
18 Box 3x3 640x480 (307200) 1.725000 1.638000
19 Dilate 3x3 640x480 (307200) 1.767000 1.663000
20 Erode 3x3 640x480 (307200) 1.765000 1.673000
21 Fast Corners No NMS 640x480 (307200) 13.685000 13.581000
22 Canny 640x480 (307200) 15.152000 15.050000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 4.705000 4.610000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 6.175000 6.076000
25 Channel Combine YUYV 640x480 (307200) 2.964000 2.866000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 2.478000 2.361000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 2.687000 2.595000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 3.931000 3.833000
29 Color Convert RGB=>RGBX 640x480 (307200) 5.295000 5.200000
30 Color Convert RGB=>NV12 640x480 (307200) 3.799000 3.701000
31 Color Convert RGB=>IYUV 640x480 (307200) 3.983000 3.883000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.925000 3.830000
33 Color Convert RGBX=>RGB 640x480 (307200) 5.048000 4.949000
34 Color Convert RGBX=>NV12 640x480 (307200) 4.668000 4.573000
35 Color Convert RGBX=>IYUV 640x480 (307200) 4.763000 4.669000
36 Color Convert RGBX=>YUV4 640x480 (307200) 4.798000 4.703000
37 Color Convert NVXX=>RGB 640x480 (307200) 3.905000 3.810000
38 Color Convert NVXX=>RGBX 640x480 (307200) 4.720000 4.624000
39 Color Convert NVXX=>IYUV 640x480 (307200) 2.402000 2.308000
40 Color Convert NVXX=>YUV4 640x480 (307200) 2.995000 2.902000
41 Color Convert YUVX=>RGB 640x480 (307200) 4.586000 4.441000
42 Color Convert YUVX=>RGBX 640x480 (307200) 5.196000 5.098000
43 Color Convert YUVX=>NV12 640x480 (307200) 3.575000 3.479000
44 Color Convert YUVX=>IYUV 640x480 (307200) 3.336000 3.242000
45 Color Convert IYUV=>RGB 640x480 (307200) 3.869000 3.774000
46 Color Convert IYUV=>RGBX 640x480 (307200) 4.715000 4.619000
47 Color Convert IYUV=>NV12 640x480 (307200) 2.480000 2.383000
48 Color Convert IYUV=>YUV4 640x480 (307200) 3.262000 3.164000
49 Convert Depth U8 to S16 640x480 (307200) 2.564000 2.467000
50 Convert Depth S16 to U8 640x480 (307200) 2.531000 2.436000
51 Convolution 3x3, O: U8 640x480 (307200) 1.959000 1.861000
52 Convolution 3x3, O: S16 640x480 (307200) 2.718000 2.621000
53 Convolution 9x3, O: U8 640x480 (307200) 18.985000 18.879000
54 Convolution 9x3, O: S16 640x480 (307200) 18.486000 18.389000
55 Convolution 3x9, O: U8 640x480 (307200) 24.636000 24.512000
56 Convolution 3x9, O: S16 640x480 (307200) 23.382000 23.279000
57 Convolution 5x5, O: U8 640x480 (307200) 2.337000 2.241000
58 Convolution 5x5, O: S16 640x480 (307200) 2.844000 2.748000
59 Convolution 7x7, O: U8 640x480 (307200) 3.184000 3.082000
60 Convolution 7x7, O: S16 640x480 (307200) 3.412000 3.320000
61 Convolution 9x9, O: U8 640x480 (307200) 37.669000 37.568000
62 Convolution 9x9, O: S16 640x480 (307200) 36.510000 36.388000
63 Equalize Histogram 640x480 (307200) 2.591000 2.497000
64 Gaussian 3x3 640x480 (307200) 1.833000 1.739000
65 Gaussian Pyramid 640x480 (307200) 10.285000 10.178000
66 Half Scale Gaussian Pyramid 640x480 (307200) 1.194000 1.102000
67 Harris Corners 640x480 (307200) 29.582000 29.469000
68 Histogram U8 640x480 (307200) 1.128000 1.033000
69 Integral Image 640x480 (307200) 4.314000 4.214000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 7.793000 7.679000
71 Laplacian Reconstruct U8 640x480 (307200) 10.413000 10.305000
72 LookUpTable S16 640x480 (307200) 3.296000 3.197000
73 LookUpTable U8 640x480 (307200) 1.845000 1.753000
74 Magnitude S16 640x480 (307200) 5.538000 5.443000
75 Mean/Standard Deviation U8 640x480 (307200) 0.926000 0.829000
76 Median 3x3 640x480 (307200) 1.884000 1.771000
77 Min Max Loc U8 640x480 (307200) 1.628000 1.532000
78 Min Max Loc S16 640x480 (307200) 2.974000 2.878000
79 Multiply U8 x U8 = U8 640x480 (307200) 2.089000 1.991000
80 Multiply U8 x U8 = S16 640x480 (307200) 2.020000 1.925000
81 Multiply U8 x S16 = S16 640x480 (307200) 4.393000 4.296000
82 Multiply S16 x S16 = S16 640x480 (307200) 3.589000 3.491000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.615000 2.519000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.606000 2.511000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 22.652000 22.551000
86 Optical Flow 5x5 640x480 (307200) 5.276000 5.172000
87 Phase 640x480 (307200) 5.147000 5.051000
88 Remap Bilinear Interpolation *640x480 (307200) 8.927000 8.827000
89 Remap Nearest Neighbor *640x480 (307200) 7.652000 7.551000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.202000 1.106000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 2.560000 2.465000
92 Sobel 3x3 640x480 (307200) 4.915000 4.816000
93 Threshold Binary 640x480 (307200) 1.674000 1.580000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.538000 0.444000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 0.833000 0.737000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.697000 0.603000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.178000 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