1 ********
2 Examples
3 ********
5 We ship three end-to-end examples within the tidl-api package
6 to demonstrate three categories of deep learning networks. The first
7 two examples can run on AM57x SoCs with either EVE or DSP devices. The last
8 example requires AM57x SoCs with both EVE and DSP. The performance
9 numbers that we present here were obtained on an AM5729 EVM, which
10 includes 2 ARM A15 cores running at 1.5GHz, 4 EVE cores at 535MHz, and
11 2 DSP cores at 750MHz.
13 For each example, we report device processing time, host processing time,
14 and TIDL API overhead. **Device processing time** is measured on the device,
15 from the moment processing starts for a frame till processing finishes.
16 **Host processing time** is measured on the host, from the moment
17 ``ProcessFrameStartAsync()`` is called till ``ProcessFrameWait()`` returns
18 in user application. It includes the TIDL API overhead, the OpenCL runtime
19 overhead, and the time to copy user input data into padded TIDL internal
20 buffers.
22 Imagenet
23 --------
25 The imagenet example takes an image as input and outputs 1000 probabilities.
26 Each probability corresponds to one object in the 1000 objects that the
27 network is pre-trained with. Our example outputs top 5 predictions
28 as the most likely objects that the input image can be.
30 The following figure and tables shows an input image, top 5 predicted
31 objects as output, and the processing time on either EVE or DSP.
33 .. image:: ../../examples/test/testvecs/input/objects/cat-pet-animal-domestic-104827.jpeg
34 :width: 600
36 .. table::
38 ==== ==============
39 Rank Object Classes
40 ==== ==============
41 1 tabby
42 2 Egyptian_cat
43 3 tiger_cat
44 4 lynx
45 5 Persian_cat
46 ==== ==============
48 .. table::
50 ====================== ==================== ============
51 Device Processing Time Host Processing Time API Overhead
52 ====================== ==================== ============
53 EVE: 123.1 ms 124.7 ms 1.34 %
54 **OR**
55 DSP: 117.9 ms 119.3 ms 1.14 %
56 ====================== ==================== ============
58 The particular network that we ran in this category, jacintonet11v2,
59 has 14 layers. User can specify whether to run the network on EVE or DSP
60 for acceleration. We can see that EVE time is slightly higher than DSP time.
61 We can also see that the overall overhead is less than 1.5%.
63 .. note::
64 The predicitions reported here are based on the output of the softmax
65 layer in the network, which are not normalized to the real probabilities.
67 Segmentation
68 ------------
70 The segmentation example takes an image as input and performs pixel-level
71 classification according to pre-trained categories. The following figures
72 show a street scene as input and the scene overlaid with pixel-level
73 classifications as output: road in green, pedestrians in red, vehicles
74 in blue and background in gray.
76 .. image:: ../../examples/test/testvecs/input/roads/pexels-photo-972355.jpeg
77 :width: 600
79 .. image:: images/pexels-photo-972355-seg.jpg
80 :width: 600
82 The network we ran in this category is jsegnet21v2, which has 26 layers.
83 From the reported time in the following table, we can see that this network
84 runs significantly faster on EVE than on DSP.
86 .. table::
88 ====================== ==================== ============
89 Device Processing Time Host Processing Time API Overhead
90 ====================== ==================== ============
91 EVE: 296.5 ms 303.3 ms 2.26 %
92 **OR**
93 DSP: 812.0 ms 818.4 ms 0.79 %
94 ====================== ==================== ============
96 .. _ssd-example:
98 SSD
99 ---
101 SSD is the abbreviation for Single Shot multi-box Detector.
102 The ssd_multibox example takes an image as input and detects multiple
103 objects with bounding boxes according to pre-trained categories.
104 The following figures show another street scene as input and the scene
105 with recognized objects boxed as output: pedestrians in red,
106 vehicles in blue and road signs in yellow.
108 .. image:: ../../examples/test/testvecs/input/roads/pexels-photo-378570.jpeg
109 :width: 600
111 .. image:: images/pexels-photo-378570-ssd.jpg
112 :width: 600
114 The network can be run entirely on either EVE or DSP. But the best
115 performance comes with running the first 30 layers on EVE and the
116 next 13 layers on DSP, for this particular jdetnet_ssd network.
117 Note the **AND** in the following table for the reported time.
118 Our end-to-end example shows how easy it is to assign a layers group id
119 to an *Executor* and how easy it is to connect the output from one
120 *ExecutionObject* to the input to another *ExecutionObject*.
122 .. table::
124 ====================== ==================== ============
125 Device Processing Time Host Processing Time API Overhead
126 ====================== ==================== ============
127 EVE: 175.2 ms 179.1 ms 2.14 %
128 **AND**
129 DSP: 21.1 ms 22.3 ms 5.62 %
130 ====================== ==================== ============
132 Test
133 ----
134 This example is used to test pre-converted networks included in the TIDL API package (``test/testvecs/config/tidl_models``). When run without any arguments, the program ``test_tidl`` will run all available networks on the C66x DSPs and EVEs available on the SoC. Use the ``-c`` option to specify a single network. Run ``test_tidl -h`` for details.
136 Running Examples
137 ----------------
139 The examples are located in ``/usr/share/ti/tidl/examples`` on
140 the EVM file system. Each example needs to be run its own directory.
141 Running an example with ``-h`` will show help message with option set.
142 The following code section shows how to run the examples, and
143 the test program that tests all supported TIDL network configs.
145 .. code:: shell
147 root@am57xx-evm:~# cd /usr/share/ti/tidl-api/examples/imagenet/
148 root@am57xx-evm:/usr/share/ti/tidl-api/examples/imagenet# make -j4
149 root@am57xx-evm:/usr/share/ti/tidl-api/examples/imagenet# ./imagenet -t d
150 Input: ../test/testvecs/input/objects/cat-pet-animal-domestic-104827.jpeg
151 frame[0]: Time on device: 117.9ms, host: 119.3ms API overhead: 1.17 %
152 1: tabby, prob = 0.996
153 2: Egyptian_cat, prob = 0.977
154 3: tiger_cat, prob = 0.973
155 4: lynx, prob = 0.941
156 5: Persian_cat, prob = 0.922
157 imagenet PASSED
159 root@am57xx-evm:/usr/share/ti/tidl-api/examples/imagenet# cd ../segmentation/; make -j4
160 root@am57xx-evm:/usr/share/ti/tidl-api/examples/segmentation# ./segmentation -i ../test/testvecs/input/roads/pexels-photo-972355.jpeg
161 Input: ../test/testvecs/input/roads/pexels-photo-972355.jpeg
162 frame[0]: Time on device: 296.5ms, host: 303.2ms API overhead: 2.21 %
163 Saving frame 0 overlayed with segmentation to: overlay_0.png
164 segmentation PASSED
166 root@am57xx-evm:/usr/share/ti/tidl-api/examples/segmentation# cd ../ssd_multibox/; make -j4
167 root@am57xx-evm:/usr/share/ti/tidl-api/examples/ssd_multibox# ./ssd_multibox -i ../test/testvecs/input/roads/pexels-photo-378570.jpeg
168 Input: ../test/testvecs/input/roads/pexels-photo-378570.jpeg
169 frame[0]: Time on EVE: 175.2ms, host: 179ms API overhead: 2.1 %
170 frame[0]: Time on DSP: 21.06ms, host: 22.43ms API overhead: 6.08 %
171 Saving frame 0 with SSD multiboxes to: multibox_0.png
172 Loop total time (including read/write/print/etc): 423.8ms
173 ssd_multibox PASSED
175 root@am57xx-evm:/usr/share/ti/tidl-api/examples/ssd_multibox# cd ../test; make -j4
176 root@am57xx-evm:/usr/share/ti/tidl-api/examples/test# ./test_tidl
177 API Version: 01.00.00.d91e442
178 Running dense_1x1 on 2 devices, type EVE
179 frame[0]: Time on device: 134.3ms, host: 135.6ms API overhead: 0.994 %
180 dense_1x1 : PASSED
181 Running j11_bn on 2 devices, type EVE
182 frame[0]: Time on device: 176.2ms, host: 177.7ms API overhead: 0.835 %
183 j11_bn : PASSED
184 Running j11_cifar on 2 devices, type EVE
185 frame[0]: Time on device: 53.86ms, host: 54.88ms API overhead: 1.85 %
186 j11_cifar : PASSED
187 Running j11_controlLayers on 2 devices, type EVE
188 frame[0]: Time on device: 122.9ms, host: 123.9ms API overhead: 0.821 %
189 j11_controlLayers : PASSED
190 Running j11_prelu on 2 devices, type EVE
191 frame[0]: Time on device: 300.8ms, host: 302.1ms API overhead: 0.437 %
192 j11_prelu : PASSED
193 Running j11_v2 on 2 devices, type EVE
194 frame[0]: Time on device: 124.1ms, host: 125.6ms API overhead: 1.18 %
195 j11_v2 : PASSED
196 Running jseg21 on 2 devices, type EVE
197 frame[0]: Time on device: 367ms, host: 374ms API overhead: 1.88 %
198 jseg21 : PASSED
199 Running jseg21_tiscapes on 2 devices, type EVE
200 frame[0]: Time on device: 302.2ms, host: 308.5ms API overhead: 2.02 %
201 frame[1]: Time on device: 301.9ms, host: 312.5ms API overhead: 3.38 %
202 frame[2]: Time on device: 302.7ms, host: 305.9ms API overhead: 1.04 %
203 frame[3]: Time on device: 301.9ms, host: 305ms API overhead: 1.01 %
204 frame[4]: Time on device: 302.7ms, host: 305.9ms API overhead: 1.05 %
205 frame[5]: Time on device: 301.9ms, host: 305.5ms API overhead: 1.17 %
206 frame[6]: Time on device: 302.7ms, host: 305.9ms API overhead: 1.06 %
207 frame[7]: Time on device: 301.9ms, host: 305ms API overhead: 1.02 %
208 frame[8]: Time on device: 297ms, host: 300.3ms API overhead: 1.09 %
209 Comparing frame: 0
210 jseg21_tiscapes : PASSED
211 Running smallRoi on 2 devices, type EVE
212 frame[0]: Time on device: 2.548ms, host: 3.637ms API overhead: 29.9 %
213 smallRoi : PASSED
214 Running squeeze1_1 on 2 devices, type EVE
215 frame[0]: Time on device: 292.9ms, host: 294.6ms API overhead: 0.552 %
216 squeeze1_1 : PASSED
218 Multiple Executor...
219 Running network tidl_config_j11_v2.txt on EVEs: 1 in thread 0
220 Running network tidl_config_j11_cifar.txt on EVEs: 0 in thread 1
221 Multiple executors: PASSED
222 Running j11_bn on 2 devices, type DSP
223 frame[0]: Time on device: 170.5ms, host: 171.5ms API overhead: 0.568 %
224 j11_bn : PASSED
225 Running j11_controlLayers on 2 devices, type DSP
226 frame[0]: Time on device: 416.4ms, host: 417.1ms API overhead: 0.176 %
227 j11_controlLayers : PASSED
228 Running j11_v2 on 2 devices, type DSP
229 frame[0]: Time on device: 118ms, host: 119.2ms API overhead: 1.01 %
230 j11_v2 : PASSED
231 Running jseg21 on 2 devices, type DSP
232 frame[0]: Time on device: 1123ms, host: 1128ms API overhead: 0.443 %
233 jseg21 : PASSED
234 Running jseg21_tiscapes on 2 devices, type DSP
235 frame[0]: Time on device: 812.3ms, host: 817.3ms API overhead: 0.614 %
236 frame[1]: Time on device: 812.6ms, host: 818.6ms API overhead: 0.738 %
237 frame[2]: Time on device: 812.3ms, host: 815.1ms API overhead: 0.343 %
238 frame[3]: Time on device: 812.7ms, host: 815.2ms API overhead: 0.312 %
239 frame[4]: Time on device: 812.3ms, host: 815.1ms API overhead: 0.353 %
240 frame[5]: Time on device: 812.6ms, host: 815.1ms API overhead: 0.302 %
241 frame[6]: Time on device: 812.2ms, host: 815.1ms API overhead: 0.357 %
242 frame[7]: Time on device: 812.6ms, host: 815.2ms API overhead: 0.315 %
243 frame[8]: Time on device: 812ms, host: 815ms API overhead: 0.367 %
244 Comparing frame: 0
245 jseg21_tiscapes : PASSED
246 Running smallRoi on 2 devices, type DSP
247 frame[0]: Time on device: 14.21ms, host: 14.94ms API overhead: 4.89 %
248 smallRoi : PASSED
249 Running squeeze1_1 on 2 devices, type DSP
250 frame[0]: Time on device: 960ms, host: 961.1ms API overhead: 0.116 %
251 squeeze1_1 : PASSED
252 tidl PASSED