TIOVX User Guide
J784S4 Linux Performance Report
SoC J784S4
Test EVM J784S4 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) 1.410000 1.234000
2 AccumulateSquare 640x480 (307200) 1.882000 1.710000
3 AccumulateWeighted 640x480 (307200) 1.007000 0.833000
4 Addition U8+U8=U8 640x480 (307200) 0.884000 0.709000
5 Addition U8+U8=S16 640x480 (307200) 0.962000 0.786000
6 Addition U8+S16=S16 640x480 (307200) 1.400000 1.232000
7 Addition S16+S16=S16 640x480 (307200) 1.478000 1.309000
8 Subtraction U8-U8=U8 640x480 (307200) 0.868000 0.703000
9 Subtraction U8-U8=S16 640x480 (307200) 0.926000 0.757000
10 Subtraction S16-U8=S16 640x480 (307200) 1.884000 1.715000
11 Subtraction S16-S16=S16 640x480 (307200) 1.450000 1.281000
12 Not Operation U8 640x480 (307200) 0.776000 0.615000
13 And Operation U8 640x480 (307200) 0.827000 0.666000
14 OR Operation U8 640x480 (307200) 0.843000 0.670000
15 XOR Operation U8 640x480 (307200) 0.832000 0.666000
16 Abs Diff U8 640x480 (307200) 0.801000 0.636000
17 Abs Diff S16 640x480 (307200) 1.803000 1.637000
18 Box 3x3 640x480 (307200) 1.482000 1.312000
19 Dilate 3x3 640x480 (307200) 0.931000 0.751000
20 Erode 3x3 640x480 (307200) 0.900000 0.736000
21 Fast Corners No NMS 640x480 (307200) 14.254000 14.064000
22 Canny 640x480 (307200) 9.224000 9.044000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 1.051000 0.876000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 1.238000 1.064000
25 Channel Combine YUYV 640x480 (307200) 1.269000 1.094000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 1.110000 0.938000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 1.975000 1.807000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 2.578000 2.398000
29 Color Convert RGB=>RGBX 640x480 (307200) 3.777000 3.595000
30 Color Convert RGB=>NV12 640x480 (307200) 3.119000 2.907000
31 Color Convert RGB=>IYUV 640x480 (307200) 3.044000 2.861000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.089000 2.907000
33 Color Convert RGBX=>RGB 640x480 (307200) 3.705000 3.518000
34 Color Convert RGBX=>NV12 640x480 (307200) 3.620000 3.437000
35 Color Convert RGBX=>IYUV 640x480 (307200) 3.551000 3.373000
36 Color Convert RGBX=>YUV4 640x480 (307200) 3.524000 3.347000
37 Color Convert NVXX=>RGB 640x480 (307200) 2.887000 2.712000
38 Color Convert NVXX=>RGBX 640x480 (307200) 3.576000 3.394000
39 Color Convert NVXX=>IYUV 640x480 (307200) 1.103000 0.937000
40 Color Convert NVXX=>YUV4 640x480 (307200) 1.289000 1.122000
41 Color Convert YUVX=>RGB 640x480 (307200) 3.400000 3.220000
42 Color Convert YUVX=>RGBX 640x480 (307200) 2.956000 2.783000
43 Color Convert YUVX=>NV12 640x480 (307200) 2.144000 1.966000
44 Color Convert YUVX=>IYUV 640x480 (307200) 2.182000 1.983000
45 Color Convert IYUV=>RGB 640x480 (307200) 2.837000 2.650000
46 Color Convert IYUV=>RGBX 640x480 (307200) 3.141000 2.964000
47 Color Convert IYUV=>NV12 640x480 (307200) 1.117000 0.942000
48 Color Convert IYUV=>YUV4 640x480 (307200) 1.131000 0.958000
49 Convert Depth U8 to S16 640x480 (307200) 0.820000 0.640000
50 Convert Depth S16 to U8 640x480 (307200) 1.385000 1.213000
51 Convolution 3x3, O: U8 640x480 (307200) 1.381000 1.208000
52 Convolution 3x3, O: S16 640x480 (307200) 1.385000 1.215000
53 Convolution 9x3, O: U8 640x480 (307200) 47.351000 47.180000
54 Convolution 9x3, O: S16 640x480 (307200) 48.280000 48.100000
55 Convolution 3x9, O: U8 640x480 (307200) 62.653000 62.479000
56 Convolution 3x9, O: S16 640x480 (307200) 63.970000 63.801000
57 Convolution 5x5, O: U8 640x480 (307200) 2.547000 2.361000
58 Convolution 5x5, O: S16 640x480 (307200) 2.561000 2.371000
59 Convolution 7x7, O: U8 640x480 (307200) 9.523000 9.355000
60 Convolution 7x7, O: S16 640x480 (307200) 4.110000 3.931000
61 Convolution 9x9, O: U8 640x480 (307200) 121.976000 121.803000
62 Convolution 9x9, O: S16 640x480 (307200) 122.925000 122.738000
63 Equalize Histogram 640x480 (307200) 2.162000 1.993000
64 Gaussian 3x3 640x480 (307200) 1.016000 0.843000
65 Gaussian Pyramid 640x480 (307200) 8.184000 8.005000
66 Half Scale Gaussian Pyramid 640x480 (307200) 0.834000 0.629000
67 Harris Corners 640x480 (307200) 17.076000 16.884000
68 Histogram U8 640x480 (307200) 1.389000 1.200000
69 Integral Image 640x480 (307200) 1.429000 1.238000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 4.038000 3.856000
71 Laplacian Reconstruct U8 640x480 (307200) 5.271000 5.097000
72 LookUpTable U8 640x480 (307200) 1.986000 1.816000
73 LookUpTable U8 640x480 (307200) 1.002000 0.835000
74 Magnitude S16 640x480 (307200) 3.757000 3.579000
75 Mean/Standard Deviation U8 640x480 (307200) 0.776000 0.613000
76 Median 3x3 640x480 (307200) 1.383000 1.212000
77 Min Max Loc U8 640x480 (307200) 1.205000 1.001000
78 Min Max Loc S16 640x480 (307200) 2.563000 2.391000
79 Multiply U8 x U8 = U8 640x480 (307200) 0.965000 0.797000
80 Multiply U8 x U8 = S16 640x480 (307200) 1.024000 0.854000
81 Multiply U8 x S16 = S16 640x480 (307200) 2.261000 2.083000
82 Multiply S16 x S16 = S16 640x480 (307200) 2.485000 2.310000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 3.045000 2.852000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 3.015000 2.839000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 60.480000 60.302000
86 Optical Flow 5x5 640x480 (307200) 4.024000 3.842000
87 Phase 640x480 (307200) 4.928000 4.743000
88 Remap Bilinear Interpolation *640x480 (307200) 9.463000 9.283000
89 Remap Nearest Neighbor *640x480 (307200) 7.550000 7.369000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.392000 1.220000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 3.777000 3.604000
92 Sobel 3x3 640x480 (307200) 1.818000 1.633000
93 Threshold Binary 640x480 (307200) 0.807000 0.625000
94 Warp Affine Nearest Neighbor *640x480 (307200) 0.438000 0.277000
95 Warp Affine Bilinear Interpolation *640x480 (307200) 1.068000 0.906000
96 Warp Pespective Nearest Neighbor *640x480 (307200) 0.643000 0.481000
97 Warp Pespective Bilinear Interpolation *640x480 (307200) 1.808000 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