TIOVX User Guide
J721S2 Linux Performance Report
SoC J721S2
Test EVM J721S2 TI EVM
HOST ARM A72 running Linux at default frequency
Target Standalone C7x DSP 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) 1.288000 1.151000
2 AccumulateSquare 640x480 (307200) 1.722000 1.607000
3 AccumulateWeighted 640x480 (307200) 0.910000 0.818000
4 Addition U8+U8=U8 640x480 (307200) 0.760000 0.640000
5 Addition U8+U8=S16 640x480 (307200) 0.867000 0.740000
6 Addition U8+S16=S16 640x480 (307200) 1.265000 1.127000
7 Addition S16+S16=S16 640x480 (307200) 1.367000 1.224000
8 Subtraction U8-U8=U8 640x480 (307200) 0.765000 0.650000
9 Subtraction U8-U8=S16 640x480 (307200) 0.854000 0.706000
10 Subtraction S16-U8=S16 640x480 (307200) 1.867000 1.728000
11 Subtraction S16-S16=S16 640x480 (307200) 1.381000 1.240000
12 Not Operation U8 640x480 (307200) 0.608000 0.504000
13 And Operation U8 640x480 (307200) 0.725000 0.611000
14 OR Operation U8 640x480 (307200) 0.722000 0.608000
15 XOR Operation U8 640x480 (307200) 0.724000 0.605000
16 Abs Diff U8 640x480 (307200) 0.759000 0.647000
17 Abs Diff S16 640x480 (307200) 1.555000 1.418000
18 Box 3x3 640x480 (307200) 1.340000 1.228000
19 Dilate 3x3 640x480 (307200) 0.692000 0.583000
20 Erode 3x3 640x480 (307200) 0.699000 0.597000
21 Fast Corners No NMS 640x480 (307200) 13.613000 13.447000
22 Canny 640x480 (307200) 7.800000 7.631000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 1.107000 0.954000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 1.323000 1.135000
25 Channel Combine YUYV 640x480 (307200) 1.068000 0.937000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 0.964000 0.837000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 1.827000 1.690000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 2.354000 2.212000
29 Color Convert RGB=>RGBX 640x480 (307200) 3.558000 3.393000
30 Color Convert RGB=>NV12 640x480 (307200) 3.126000 2.985000
31 Color Convert RGB=>IYUV 640x480 (307200) 3.066000 2.919000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.120000 2.974000
33 Color Convert RGBX=>RGB 640x480 (307200) 3.596000 3.429000
34 Color Convert RGBX=>NV12 640x480 (307200) 3.625000 3.483000
35 Color Convert RGBX=>IYUV 640x480 (307200) 3.593000 3.442000
36 Color Convert RGBX=>YUV4 640x480 (307200) 3.620000 3.463000
37 Color Convert NVXX=>RGB 640x480 (307200) 2.849000 2.696000
38 Color Convert NVXX=>RGBX 640x480 (307200) 3.625000 3.477000
39 Color Convert NVXX=>IYUV 640x480 (307200) 1.004000 0.867000
40 Color Convert NVXX=>YUV4 640x480 (307200) 1.126000 0.991000
41 Color Convert YUVX=>RGB 640x480 (307200) 3.188000 3.037000
42 Color Convert YUVX=>RGBX 640x480 (307200) 2.849000 2.694000
43 Color Convert YUVX=>NV12 640x480 (307200) 1.884000 1.751000
44 Color Convert YUVX=>IYUV 640x480 (307200) 1.879000 1.751000
45 Color Convert IYUV=>RGB 640x480 (307200) 2.905000 2.762000
46 Color Convert IYUV=>RGBX 640x480 (307200) 3.038000 2.889000
47 Color Convert IYUV=>NV12 640x480 (307200) 1.016000 0.886000
48 Color Convert IYUV=>YUV4 640x480 (307200) 1.035000 0.895000
49 Convert Depth U8 to S16 640x480 (307200) 0.638000 0.518000
50 Convert Depth S16 to U8 640x480 (307200) 1.219000 1.099000
51 Convolution 3x3, O: U8 640x480 (307200) 1.255000 1.145000
52 Convolution 3x3, O: S16 640x480 (307200) 1.194000 1.083000
53 Convolution 9x3, O: U8 640x480 (307200) 47.335000 47.222000
54 Convolution 9x3, O: S16 640x480 (307200) 48.277000 48.147000
55 Convolution 3x9, O: U8 640x480 (307200) 62.568000 62.455000
56 Convolution 3x9, O: S16 640x480 (307200) 64.019000 63.898000
57 Convolution 5x5, O: U8 640x480 (307200) 2.427000 2.236000
58 Convolution 5x5, O: S16 640x480 (307200) 2.354000 2.230000
59 Convolution 7x7, O: U8 640x480 (307200) 3.731000 3.616000
60 Convolution 7x7, O: S16 640x480 (307200) 4.105000 3.991000
61 Convolution 9x9, O: U8 640x480 (307200) 121.971000 121.834000
62 Convolution 9x9, O: S16 640x480 (307200) 122.889000 122.760000
63 Equalize Histogram 640x480 (307200) 1.972000 1.851000
64 Gaussian 3x3 640x480 (307200) 0.851000 0.749000
65 Gaussian Pyramid 640x480 (307200) 7.620000 7.479000
66 Half Scale Gaussian Pyramid 640x480 (307200) 0.673000 0.572000
67 Harris Corners 640x480 (307200) 17.580000 17.392000
68 Histogram U8 640x480 (307200) 1.205000 1.113000
69 Integral Image 640x480 (307200) 1.271000 1.121000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 3.661000 3.504000
71 Laplacian Reconstruct U8 640x480 (307200) 4.613000 4.453000
72 LookUpTable U8 640x480 (307200) 1.965000 1.825000
73 LookUpTable U8 640x480 (307200) 0.829000 0.719000
74 Magnitude S16 640x480 (307200) 3.867000 3.716000
75 Mean/Standard Deviation U8 640x480 (307200) 0.541000 0.448000
76 Median 3x3 640x480 (307200) 1.288000 1.166000
77 Min Max Loc U8 640x480 (307200) 0.963000 0.869000
78 Min Max Loc S16 640x480 (307200) 1.937000 1.833000
79 Multiply U8 x U8 = U8 640x480 (307200) 0.850000 0.731000
80 Multiply U8 x U8 = S16 640x480 (307200) 0.932000 0.808000
81 Multiply U8 x S16 = S16 640x480 (307200) 2.016000 1.876000
82 Multiply S16 x S16 = S16 640x480 (307200) 2.493000 2.344000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.789000 2.678000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.798000 2.682000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 60.286000 60.166000
86 Optical Flow 5x5 640x480 (307200) 3.755000 3.601000
87 Phase 640x480 (307200) 4.865000 4.723000
88 Remap Bilinear Interpolation *640x480 (307200) 9.749000 9.579000
89 Remap Nearest Neighbor *640x480 (307200) 8.591000 8.428000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.304000 1.199000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 5.195000 5.089000
92 Sobel 3x3 640x480 (307200) 1.562000 1.414000
93 Threshold Binary 640x480 (307200) 0.634000 0.521000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.361000 0.267000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 1.133000 1.040000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.605000 0.513000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.738000 1.641000

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