User Manual

6 AI accelerator card
multiple_tpus_performance_analysis.cc:108] Model name:
mobilenet_v1_1.0_224_quant_edgetpu.tite
multiple_tpus_performance_analysis.cc:82] Single Edge TPU base
time 69760.4 seconds.
multiple_tpus_performance_analysis.cc:84] # TPUs: speedup:
1.99319
multiple_tpus_performance_analysis.cc:84] # TPUs: speedup:
2.9895
multiple_tpus_performance_analysis.cc:84] # TPUs: speedup:
3.94866
multiple_tpus_performance_analysis.cc:84] # TPUs: speedup:
4.89411
multiple_tpus_performance_analysis.cc:84] # TPUs: speedup:
5.77968
multiple_tpus_performance_analysis.cc:84] # TPUs: speedup:
6.66029
3. You can check the test result through the Pass or fail log.
2.3 Begin multiple Edge TPU tests
1. Clone the "edgetpu" source code of Google from github.
$ git clone https://github.com/google-coral/edgetpu
$ cd edgetpu/benchmarks
$ python3 multiple_tpus_performance_analysis.py
2. Open the “edgetpu” folder, then enter the following commands to test
multiple Edge TPUs: