Installing the TPU driver
The official guide only covers steps for Debian. Third-party compiled binaries are available for Fedora, RHEL, and OpenSUSE etc. The RPM for OpenSUSE Tumbleweed can be downloaded directly here.
Once downloaded, install the RPM file with "sudo zypper install gasket..."
Seems OpenSUSE now has the gasket driver in repository. Just install it with "sudo zypper install gasket-driver gasket-driver-kmp-default"
Installing the userspace runtime driver
The userspace driver can be built from
source inside docker:
git clone https://github.com/google-coral/libedgetpu.git
cd libedgetpu
CPU=k8 DOCKER_CPUS="k8" DOCKER_IMAGE="ubuntu:18.04" DOCKER_TARGETS=libedgetpu make docker-build
Copy the files from "out" folder and install using ldconfig:
sudo mkdir -p /usr/lib/tpu/lib
sudo cp --preserve=links out/direct/k8/* /usr/lib/tpu/lib
echo /usr/lib/tpu/lib | sudo tee -a /etc/ld.so.conf.d/tpu.conf
sudo ldconfig
Grant permission
Running the Python examples may fail with error "failed to load delegate...". This is due to permission issue. Add your user to the group "apex" with the command "sudo usermod -aG apex [your username]".
Also, if using the USB accelerator instead, add two rules to the udev file "/etc/udev/rules.d/65-apex.rules" and reboot:
SUBSYSTEM=="usb",ATTRS{idVendor}=="1a6e",GROUP="apex"
SUBSYSTEM=="usb",ATTRS{idVendor}=="18d1",GROUP="apex"
Running the examples
If you followed the above steps to install drivers, you may need to reboot before running any example.
The "pycoral" package mentioned on the officail guide is kind of confusing. The "pycoral" from pypi is actually a totally unrelated package. What we need for the TPU can be found
here.
Also, there is no support for Python 3.10 yet. So needed to use 3.9. To install packages in order to run the example:
python3.9 -m venv venv
. venv/bin/activate
pip install numpy pillow
python -m pip install --extra-index-url https://google-coral.github.io/py-repo/ pycoral~=2.0
Then follow the steps on the official guide to run the example.
$ python3 examples/classify_image.py --model test_data/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite --labels test_data/inat_bird_labels.txt --input test_data/parrot.jpg
----INFERENCE TIME----
Note: The first inference on Edge TPU is slow because it includes loading the model into Edge TPU memory.
12.8ms
2.8ms
2.8ms
2.8ms
2.8ms
-------RESULTS--------
Ara macao (Scarlet Macaw): 0.75781
