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) 2.643000 2.548000
2 AccumulateSquare 640x480 (307200) 2.673000 2.579000
3 AccumulateWeighted 640x480 (307200) 1.756000 1.660000
4 Addition U8+U8=U8 640x480 (307200) 2.089000 1.994000
5 Addition U8+U8=S16 640x480 (307200) 3.129000 3.032000
6 Addition U8+S16=S16 640x480 (307200) 4.275000 4.178000
7 Addition S16+S16=S16 640x480 (307200) 3.609000 3.512000
8 Subtraction U8-U8=U8 640x480 (307200) 2.586000 2.490000
9 Subtraction U8-U8=S16 640x480 (307200) 3.110000 3.013000
10 Subtraction S16-U8=S16 640x480 (307200) 3.871000 3.742000
11 Subtraction S16-S16=S16 640x480 (307200) 4.557000 4.462000
12 Not Operation U8 640x480 (307200) 1.573000 1.478000
13 And Operation U8 640x480 (307200) 2.051000 1.955000
14 OR Operation U8 640x480 (307200) 2.036000 1.937000
15 XOR Operation U8 640x480 (307200) 2.033000 1.938000
16 Abs Diff U8 640x480 (307200) 2.599000 2.505000
17 Abs Diff S16 640x480 (307200) 4.508000 4.409000
18 Box 3x3 640x480 (307200) 1.720000 1.626000
19 Dilate 3x3 640x480 (307200) 1.767000 1.673000
20 Erode 3x3 640x480 (307200) 1.769000 1.672000
21 Fast Corners No NMS 640x480 (307200) 13.711000 13.607000
22 Canny 640x480 (307200) 15.172000 15.072000
23 Channel Combine 3 to 1 (RGB) 640x480 (307200) 4.704000 4.602000
24 Channel Combine 4 to 1 (RGBX) 640x480 (307200) 6.179000 6.079000
25 Channel Combine YUYV 640x480 (307200) 2.962000 2.862000
26 Channel Combine 2 to 1 (NV12) 640x480 (307200) 2.442000 2.347000
27 Channel Extract 1 of 3 (RGB) 640x480 (307200) 2.693000 2.595000
28 Channel Extract 1 of 4 (RGBX) 640x480 (307200) 3.932000 3.835000
29 Color Convert RGB=>RGBX 640x480 (307200) 5.336000 5.241000
30 Color Convert RGB=>NV12 640x480 (307200) 3.802000 3.708000
31 Color Convert RGB=>IYUV 640x480 (307200) 3.996000 3.892000
32 Color Convert RGB=>YUV4 640x480 (307200) 3.916000 3.819000
33 Color Convert RGBX=>RGB 640x480 (307200) 5.040000 4.945000
34 Color Convert RGBX=>NV12 640x480 (307200) 4.667000 4.571000
35 Color Convert RGBX=>IYUV 640x480 (307200) 4.779000 4.683000
36 Color Convert RGBX=>YUV4 640x480 (307200) 4.792000 4.694000
37 Color Convert NVXX=>RGB 640x480 (307200) 3.900000 3.806000
38 Color Convert NVXX=>RGBX 640x480 (307200) 4.733000 4.633000
39 Color Convert NVXX=>IYUV 640x480 (307200) 2.376000 2.286000
40 Color Convert NVXX=>YUV4 640x480 (307200) 2.998000 2.900000
41 Color Convert YUVX=>RGB 640x480 (307200) 4.503000 4.406000
42 Color Convert YUVX=>RGBX 640x480 (307200) 5.240000 5.140000
43 Color Convert YUVX=>NV12 640x480 (307200) 3.586000 3.489000
44 Color Convert YUVX=>IYUV 640x480 (307200) 3.347000 3.248000
45 Color Convert IYUV=>RGB 640x480 (307200) 3.811000 3.715000
46 Color Convert IYUV=>RGBX 640x480 (307200) 4.661000 4.561000
47 Color Convert IYUV=>NV12 640x480 (307200) 2.438000 2.343000
48 Color Convert IYUV=>YUV4 640x480 (307200) 3.226000 3.128000
49 Convert Depth U8 to S16 640x480 (307200) 2.594000 2.497000
50 Convert Depth S16 to U8 640x480 (307200) 2.532000 2.438000
51 Convolution 3x3, O: U8 640x480 (307200) 1.989000 1.883000
52 Convolution 3x3, O: S16 640x480 (307200) 2.724000 2.629000
53 Convolution 9x3, O: U8 640x480 (307200) 18.993000 18.891000
54 Convolution 9x3, O: S16 640x480 (307200) 18.476000 18.381000
55 Convolution 3x9, O: U8 640x480 (307200) 24.588000 24.485000
56 Convolution 3x9, O: S16 640x480 (307200) 23.397000 23.296000
57 Convolution 5x5, O: U8 640x480 (307200) 2.355000 2.251000
58 Convolution 5x5, O: S16 640x480 (307200) 2.854000 2.757000
59 Convolution 7x7, O: U8 640x480 (307200) 3.198000 3.103000
60 Convolution 7x7, O: S16 640x480 (307200) 3.438000 3.342000
61 Convolution 9x9, O: U8 640x480 (307200) 37.672000 37.572000
62 Convolution 9x9, O: S16 640x480 (307200) 36.478000 36.378000
63 Equalize Histogram 640x480 (307200) 2.592000 2.496000
64 Gaussian 3x3 640x480 (307200) 1.862000 1.754000
65 Gaussian Pyramid 640x480 (307200) 10.295000 10.187000
66 Half Scale Gaussian Pyramid 640x480 (307200) 1.211000 1.111000
67 Harris Corners 640x480 (307200) 29.582000 29.476000
68 Histogram U8 640x480 (307200) 1.129000 1.032000
69 Integral Image 640x480 (307200) 4.325000 4.228000
70 Laplacian Pyramid U8; Levels = 6 640x480 (307200) 7.790000 7.677000
71 Laplacian Reconstruct U8 640x480 (307200) 10.428000 10.319000
72 LookUpTable S16 640x480 (307200) 3.294000 3.201000
73 LookUpTable U8 640x480 (307200) 1.844000 1.749000
74 Magnitude S16 640x480 (307200) 5.560000 5.458000
75 Mean/Standard Deviation U8 640x480 (307200) 0.937000 0.836000
76 Median 3x3 640x480 (307200) 1.874000 1.776000
77 Min Max Loc U8 640x480 (307200) 1.578000 1.482000
78 Min Max Loc S16 640x480 (307200) 2.968000 2.870000
79 Multiply U8 x U8 = U8 640x480 (307200) 2.095000 1.993000
80 Multiply U8 x U8 = S16 640x480 (307200) 2.020000 1.920000
81 Multiply U8 x S16 = S16 640x480 (307200) 4.371000 4.275000
82 Multiply S16 x S16 = S16 640x480 (307200) 3.578000 3.476000
83 NonLinear Filter mask=5x5; Erosion 640x480 (307200) 2.631000 2.531000
84 NonLinear Filter mask=5x5; Dilation 640x480 (307200) 2.613000 2.511000
85 NonLinear Filter mask=5x5; Median 640x480 (307200) 22.698000 22.594000
86 Optical Flow 5x5 640x480 (307200) 5.282000 5.171000
87 Phase 640x480 (307200) 5.153000 5.055000
88 Remap Bilinear Interpolation *640x480 (307200) 8.899000 8.795000
89 Remap Nearest Neighbor *640x480 (307200) 7.735000 7.630000
90 Scale Image 1 to 2; Nearest Neighbor *640x480 (307200) 1.211000 1.115000
91 Scale Image 1 to 2; Bilinear Interpolation *640x480 (307200) 2.564000 2.467000
92 Sobel 3x3 640x480 (307200) 4.930000 4.837000
93 Threshold Binary 640x480 (307200) 1.679000 1.578000
94 Warp Affine Nearest Neighbor *256x256 (65536) 0.529000 0.440000
95 Warp Affine Bilinear Interpolation *256x256 (65536) 0.853000 0.755000
96 Warp Pespective Nearest Neighbor *256x256 (65536) 0.716000 0.621000
97 Warp Pespective Bilinear Interpolation *256x256 (65536) 1.188000 1.092000

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