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) 1.327000 1.224000
2 AccumulateSquare 640x480 (307200) 1.940000 1.849000
3 AccumulateWeighted 640x480 (307200) 0.917000 0.846000
4 Addition U8+U8=U8 640x480 (307200) 0.800000 0.726000
5 Addition U8+U8=S16 640x480 (307200) 0.827000 0.751000
6 Addition U8+S16=S16 640x480 (307200) 1.406000 1.316000
7 Addition S16+S16=S16 640x480 (307200) 1.463000 1.389000
8 Subtraction U8-U8=U8 640x480 (307200) 0.796000 0.723000
9 Subtraction U8-U8=S16 640x480 (307200) 0.806000 0.734000
10 Subtraction S16-U8=S16 640x480 (307200) 1.905000 1.828000
11 Subtraction S16-S16=S16 640x480 (307200) 1.443000 1.366000
12 Not Operation U8 640x480 (307200) 0.615000 0.546000
13 And Operation U8 640x480 (307200) 0.761000 0.690000
14 OR Operation U8 640x480 (307200) 0.750000 0.680000
15 XOR Operation U8 640x480 (307200) 0.752000 0.681000
16 Abs Diff U8 640x480 (307200) 0.788000 0.719000
17 Abs Diff S16 640x480 (307200) 1.685000 1.611000
18 Box 3x3 640x480 (307200) 1.457000 1.389000
19 Dilate 3x3 640x480 (307200) 0.634000 0.562000
20 Erode 3x3 640x480 (307200) 0.629000 0.557000
21 Fast Corners No NMS 640x480 (307200) 13.485000 13.392000
22 Canny 640x480 (307200) 7.650000 7.565000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 1.049000 0.964000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 1.137000 1.055000
25 Channel Combine YUYV 640x480 (307200) 0.931000 0.859000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 0.833000 0.735000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 1.660000 1.586000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 2.231000 2.158000
29 Color Convert RGB=>RGBX 640x480 (307200) 3.522000 3.440000
30 Color Convert RGB=>NV12 640x480 (307200) 3.026000 2.949000
31 Color Convert RGB=>IYUV 640x480 (307200) 2.984000 2.909000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.131000 3.048000
33 Color Convert RGBX=>RGB 640x480 (307200) 3.783000 3.703000
34 Color Convert RGBX=>NV12 640x480 (307200) 3.559000 3.482000
35 Color Convert RGBX=>IYUV 640x480 (307200) 3.525000 3.447000
36 Color Convert RGBX=>YUV4 640x480 (307200) 3.641000 3.560000
37 Color Convert NVXX=>RGB 640x480 (307200) 2.858000 2.777000
38 Color Convert NVXX=>RGBX 640x480 (307200) 3.631000 3.549000
39 Color Convert NVXX=>IYUV 640x480 (307200) 0.799000 0.719000
40 Color Convert NVXX=>YUV4 640x480 (307200) 1.036000 0.957000
41 Color Convert YUVX=>RGB 640x480 (307200) 3.254000 3.174000
42 Color Convert YUVX=>RGBX 640x480 (307200) 3.018000 2.939000
43 Color Convert YUVX=>NV12 640x480 (307200) 1.786000 1.707000
44 Color Convert YUVX=>IYUV 640x480 (307200) 1.783000 1.709000
45 Color Convert IYUV=>RGB 640x480 (307200) 2.931000 2.845000
46 Color Convert IYUV=>RGBX 640x480 (307200) 3.153000 3.069000
47 Color Convert IYUV=>NV12 640x480 (307200) 0.822000 0.744000
48 Color Convert IYUV=>YUV4 640x480 (307200) 0.888000 0.813000
49 Convert Depth U8 to S16 640x480 (307200) 0.545000 0.469000
50 Convert Depth S16 to U8 640x480 (307200) 1.040000 0.955000
51 Convolution 3x3, O: U8 640x480 (307200) 1.165000 1.092000
52 Convolution 3x3, O: S16 640x480 (307200) 1.105000 1.030000
53 Convolution 9x3, O: U8 640x480 (307200) 47.232000 47.147000
54 Convolution 9x3, O: S16 640x480 (307200) 48.131000 48.050000
55 Convolution 3x9, O: U8 640x480 (307200) 62.465000 62.388000
56 Convolution 3x9, O: S16 640x480 (307200) 63.851000 63.772000
57 Convolution 5x5, O: U8 640x480 (307200) 2.210000 2.137000
58 Convolution 5x5, O: S16 640x480 (307200) 2.223000 2.152000
59 Convolution 7x7, O: U8 640x480 (307200) 3.634000 3.561000
60 Convolution 7x7, O: S16 640x480 (307200) 4.055000 3.979000
61 Convolution 9x9, O: U8 640x480 (307200) 121.857000 121.776000
62 Convolution 9x9, O: S16 640x480 (307200) 122.764000 122.686000
63 Equalize Histogram 640x480 (307200) 1.991000 1.919000
64 Gaussian 3x3 640x480 (307200) 0.809000 0.738000
65 Gaussian Pyramid 640x480 (307200) 6.889000 6.801000
66 Half Scale Gaussian Pyramid 640x480 (307200) 0.598000 0.528000
67 Harris Corners 640x480 (307200) 16.488000 16.397000
68 Histogram U8 640x480 (307200) 1.129000 1.056000
69 Integral Image 640x480 (307200) 1.116000 1.038000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 3.181000 3.100000
71 Laplacian Reconstruct U8 640x480 (307200) 4.324000 4.244000
72 LookUpTable U8 640x480 (307200) 1.718000 1.639000
73 LookUpTable U8 640x480 (307200) 0.762000 0.691000
74 Magnitude S16 640x480 (307200) 3.948000 3.870000
75 Mean/Standard Deviation U8 640x480 (307200) 0.514000 0.445000
76 Median 3x3 640x480 (307200) 1.291000 1.217000
77 Min Max Loc U8 640x480 (307200) 0.949000 0.878000
78 Min Max Loc S16 640x480 (307200) 1.910000 1.831000
79 Multiply U8 x U8 = U8 640x480 (307200) 0.822000 0.750000
80 Multiply U8 x U8 = S16 640x480 (307200) 0.907000 0.832000
81 Multiply U8 x S16 = S16 640x480 (307200) 2.180000 2.104000
82 Multiply S16 x S16 = S16 640x480 (307200) 2.692000 2.616000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.679000 2.608000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.677000 2.606000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 60.118000 60.043000
86 Optical Flow 5x5 640x480 (307200) 4.941000 4.856000
87 Phase 640x480 (307200) 4.710000 4.632000
88 Remap Bilinear Interpolation *640x480 (307200) 9.012000 8.929000
89 Remap Nearest Neighbor *640x480 (307200) 7.489000 7.403000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.346000 1.274000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 5.159000 5.090000
92 Sobel 3x3 640x480 (307200) 1.449000 1.365000
93 Threshold Binary 640x480 (307200) 0.565000 0.486000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.334000 0.261000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 1.087000 1.014000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.551000 0.482000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.687000 1.618000

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