Introduction
This download is primarily intended for users running a bare-metal version of Ubuntu on a PC with an Intel CPU (and integrated GPU). The iGPU on your CPU will be used for both video decoding and the use of analytics.
You can use a virtual machine (especially if you're not using analytics); however, using the analytics will fall back to CPU-only if one isn't available (and not properly configured using the instructions below). In these circumstances, the performance of analytics will be considerably worse, and you should use the encoder's Performance page to determine if your container is overloaded.
Download
This preview release is only available as a .DEB installation file.
Pre-Installation - setting up your iGPU
The analytics will use the Intel GPU if you have one available. This involves installing some other libraries before installation:
sudo apt install intel-opencl-icd clinfo intel-gpu-tools
To test the GPU you'll need to grant yourself access to the GPU (perform this logged in using your user account):
sudo usermod -a -G video,render <your username>
# make sure user is adding to groups in current shell
newgrp video
newgrp render
You can then see if the Intel GPU is recognised using clinfo:
clinfo -l
Platform #0: Intel(R) OpenCL Graphics
`-- Device #0: Intel(R) UHD Graphics 770
If you see no output, then the GPU is not working, and using analytics will be CPU-only and use considerably more resources.
Installation
This version will not work if you upgrade from an earlier version! It is also not upgradable, and should be uninstalled once testing is complete
If you already have any version of EVVR Container installed, you must uninstall it:
sudo apt remove --purge edgevis-encoder
Install the current version of the DEB file (replacing XX with the version number in your downloaded file):
sudo apt install ./edgevis-analytics_9.3.90-TCXX_amd64_cpu.deb
The next step is to optimise the new analytics:
sudo edgevis-encoder-analytics --prepare
You can, if desired, customise your installation using configuration options in your container's compose file.
Log in to your container's web interface using the usual mechanism.