Trying to understand the software suite or DFC requirements and installation

My setup is as follows:

  • Dev environment : MacOS
  • Hailo hardware: 8 running on raspberry pi 5

I’m trying to install the DFC or SW suite but the requirements are not clear. If I try to install in a ubuntu docker container running on my mac will that work?

Do I have to have the hailo8 installed on the machine i’m installing it on?

There is a separate set of requirements for nvidia for hardware emulation. I can’t see what this is referring to. Can I do everything I need without this.

I’ve been trying for 2 days to install the DFC on a docker container on my mac and I’m not sure it’s a hardware issue or my internet connection (I’m currently away from home and my network speed is the worst I’ve ever seen) The first issue I have is when it get’s to the tensorflow-2.12.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl step it frequently hits ReadTimeoutError(self._pool, None, "Read timed out.") because the connection is so shabby. But the other issue I keep facing is that many of the packages it tries to install hit this issue:

Downloading types_python_dateutil-2.9.0.20241206-py3-none-any.whl (14 kB)
ERROR: THESE PACKAGES DO NOT MATCH THE HASHES FROM THE REQUIREMENTS FILE. If you have updated the package versions, please update the hashes. Otherwise, examine the package contents carefully; someone may have tampered with them.
    unknown package:
        Expected sha256 357d9d2851188a8d27ee195345b4d175cad970150d1344ba9d9fcc4bf2b68336
             Got        06c834bcbf503ac1848afc2e2d70543e8dfffa42783d6487cbf74005d136e78f

So I have to find a work around and start all over again (with my slow internet).

So in summary:

  1. Can the host machine for docker be a Mac?
  2. Do I need an nvidia GPU (if so what for)?
  3. Do I need the hailo chip to be connected to the host?
  4. What are these HASH issues being caused by when installing?

Hi @rosslote,

  1. Can the host machine for docker be a Mac?
    The Hailo AI SW Suite or the DFC require an Ubuntu 20/22 host machine. The docker itself has been tested an validated for those OS only. A usage under WSL is possible - with some limitations - but no support is provided for MAC OS for our docker.
    For the same reason, we cannot state that the DFC can run in your own custom docker (Ubuntu) running on a MAC OS host machine
  2. Do I need an nvidia GPU (if so what for)?
    The GPU is needed only to run advanced optimization algorithms, or to speed up the execution time of the emulator (if you want to simulate the model on the PC after the optimization phase). Basic optimization does not require a GPU, but we suggest to have one available in the system, in case you have to run IBC/Finetuning or Adaround
  3. Do I need the hailo chip to be connected to the host?
    Only if you want to run inference on it. This is not necessary for model conversion.
  4. What are these HASH issues being caused by when installing?
    Not sure about these errors. Since you are using your own Docker and you are installing the DFC inside it, it could be related to how the Docker was set up, or a network issue.