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) 2.693000 2.595000
2 AccumulateSquare 640x480 (307200) 2.687000 2.604000
3 AccumulateWeighted 640x480 (307200) 1.732000 1.650000
4 Addition U8+U8=U8 640x480 (307200) 2.077000 1.992000
5 Addition U8+U8=S16 640x480 (307200) 3.047000 2.965000
6 Addition U8+S16=S16 640x480 (307200) 4.168000 4.080000
7 Addition S16+S16=S16 640x480 (307200) 3.773000 3.688000
8 Subtraction U8-U8=U8 640x480 (307200) 2.567000 2.480000
9 Subtraction U8-U8=S16 640x480 (307200) 3.074000 2.986000
10 Subtraction S16-U8=S16 640x480 (307200) 3.823000 3.736000
11 Subtraction S16-S16=S16 640x480 (307200) 4.865000 4.781000
12 Not Operation U8 640x480 (307200) 1.551000 1.468000
13 And Operation U8 640x480 (307200) 2.034000 1.947000
14 OR Operation U8 640x480 (307200) 2.013000 1.927000
15 XOR Operation U8 640x480 (307200) 2.022000 1.940000
16 Abs Diff U8 640x480 (307200) 2.511000 2.428000
17 Abs Diff S16 640x480 (307200) 4.813000 4.733000
18 Box 3x3 640x480 (307200) 1.714000 1.634000
19 Dilate 3x3 640x480 (307200) 1.676000 1.594000
20 Erode 3x3 640x480 (307200) 1.680000 1.596000
21 Fast Corners No NMS 640x480 (307200) 13.486000 13.396000
22 Canny 640x480 (307200) 15.023000 14.927000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 4.608000 4.523000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 5.946000 5.858000
25 Channel Combine YUYV 640x480 (307200) 2.909000 2.821000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 2.390000 2.302000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 2.674000 2.593000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 3.895000 3.813000
29 Color Convert RGB=>RGBX 640x480 (307200) 5.276000 5.189000
30 Color Convert RGB=>NV12 640x480 (307200) 3.846000 3.760000
31 Color Convert RGB=>IYUV 640x480 (307200) 3.972000 3.886000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.973000 3.888000
33 Color Convert RGBX=>RGB 640x480 (307200) 5.020000 4.934000
34 Color Convert RGBX=>NV12 640x480 (307200) 4.647000 4.558000
35 Color Convert RGBX=>IYUV 640x480 (307200) 4.676000 4.587000
36 Color Convert RGBX=>YUV4 640x480 (307200) 4.752000 4.667000
37 Color Convert NVXX=>RGB 640x480 (307200) 3.770000 3.683000
38 Color Convert NVXX=>RGBX 640x480 (307200) 4.688000 4.600000
39 Color Convert NVXX=>IYUV 640x480 (307200) 2.359000 2.273000
40 Color Convert NVXX=>YUV4 640x480 (307200) 3.024000 2.940000
41 Color Convert YUVX=>RGB 640x480 (307200) 4.449000 4.364000
42 Color Convert YUVX=>RGBX 640x480 (307200) 5.308000 5.220000
43 Color Convert YUVX=>NV12 640x480 (307200) 3.546000 3.462000
44 Color Convert YUVX=>IYUV 640x480 (307200) 3.303000 3.215000
45 Color Convert IYUV=>RGB 640x480 (307200) 3.917000 3.833000
46 Color Convert IYUV=>RGBX 640x480 (307200) 4.687000 4.602000
47 Color Convert IYUV=>NV12 640x480 (307200) 2.415000 2.330000
48 Color Convert IYUV=>YUV4 640x480 (307200) 3.298000 3.210000
49 Convert Depth U8 to S16 640x480 (307200) 2.409000 2.323000
50 Convert Depth S16 to U8 640x480 (307200) 2.523000 2.435000
51 Convolution 3x3, O: U8 640x480 (307200) 1.882000 1.796000
52 Convolution 3x3, O: S16 640x480 (307200) 2.609000 2.525000
53 Convolution 9x3, O: U8 640x480 (307200) 18.923000 18.836000
54 Convolution 9x3, O: S16 640x480 (307200) 18.378000 18.290000
55 Convolution 3x9, O: U8 640x480 (307200) 24.507000 24.420000
56 Convolution 3x9, O: S16 640x480 (307200) 23.283000 23.194000
57 Convolution 5x5, O: U8 640x480 (307200) 2.223000 2.139000
58 Convolution 5x5, O: S16 640x480 (307200) 2.737000 2.654000
59 Convolution 7x7, O: U8 640x480 (307200) 3.079000 2.996000
60 Convolution 7x7, O: S16 640x480 (307200) 3.301000 3.214000
61 Convolution 9x9, O: U8 640x480 (307200) 37.587000 37.499000
62 Convolution 9x9, O: S16 640x480 (307200) 36.387000 36.293000
63 Equalize Histogram 640x480 (307200) 2.572000 2.488000
64 Gaussian 3x3 640x480 (307200) 1.742000 1.659000
65 Gaussian Pyramid 640x480 (307200) 10.511000 10.413000
66 Half Scale Gaussian Pyramid 640x480 (307200) 1.132000 1.049000
67 Harris Corners 640x480 (307200) 29.337000 29.243000
68 Histogram U8 640x480 (307200) 1.083000 1.000000
69 Integral Image 640x480 (307200) 4.192000 4.108000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 7.699000 7.595000
71 Laplacian Reconstruct U8 640x480 (307200) 10.524000 10.427000
72 LookUpTable U8 640x480 (307200) 3.264000 3.177000
73 LookUpTable U8 640x480 (307200) 1.732000 1.649000
74 Magnitude S16 640x480 (307200) 5.732000 5.645000
75 Mean/Standard Deviation U8 640x480 (307200) 0.899000 0.814000
76 Median 3x3 640x480 (307200) 1.799000 1.714000
77 Min Max Loc U8 640x480 (307200) 1.620000 1.523000
78 Min Max Loc S16 640x480 (307200) 3.004000 2.915000
79 Multiply U8 x U8 = U8 640x480 (307200) 2.087000 2.000000
80 Multiply U8 x U8 = S16 640x480 (307200) 2.025000 1.937000
81 Multiply U8 x S16 = S16 640x480 (307200) 4.292000 4.206000
82 Multiply S16 x S16 = S16 640x480 (307200) 3.805000 3.716000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.560000 2.475000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.543000 2.458000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 22.612000 22.519000
86 Optical Flow 5x5 640x480 (307200) 5.310000 5.217000
87 Phase 640x480 (307200) 5.027000 4.938000
88 Remap Bilinear Interpolation *640x480 (307200) 8.888000 8.798000
89 Remap Nearest Neighbor *640x480 (307200) 7.596000 7.507000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.152000 1.065000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 2.521000 2.433000
92 Sobel 3x3 640x480 (307200) 4.838000 4.748000
93 Threshold Binary 640x480 (307200) 1.659000 1.572000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.520000 0.434000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 0.851000 0.766000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.653000 0.567000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.133000 1.045000

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