On 02/01/2025 12:56, Sakthivel Velumani wrote:

You should be able to run UHD-using applications on the ARM on the X410,
just like any of the other platforms that provide
  an embedded ARM CPU.  HOWEVER, those ARMs are only dual core,
typically (I don't know for sure on the X410), and they
  cannot be expected to provide high-bandwidth sample streams
into/out-of the radio.

I don't know whether the Python API for UHD is installed by default on
the X410 system image, but that would be an
  easy pathway to do initial experiments.

Really though, for these high-end boxes, the ARM CPUs exist largely for
housekeeping and logistics management of
  the radio, via things like MPM, etc.   The fact that they can run a
user application is just kind of a side-effect, and
  that application would need to offload nearly-ALL of its actual DSP
work to RFNoC, since these ARM CPUs are not
  particularly capable in terms of computational and memory-bandwidth
performance.

Hi Marcus,

Thanks for the pointers. I am looking to offload all the DSP work to the
RFNoC and have the CPUs do trivial tasks like sequence generation so
that I don't have to write HDL for everything. I am still not entirely
sure of how this can be done but I found a software development guide
for E310
(https://kb.ettus.com/Software_Development_on_the_E310_and_E312) which
looks like something I want to achieve so it's a good start I guess.

Thanks again.

Sakthi

On 1/2/25 13:10, Marcus D. Leech wrote:

On 02/01/2025 13:56, Sakthivel Velumani wrote:

I think that the Python binding library for UHD IS installed in the
system image on the X410.

Go into Python3 and

import uhd

That should work.

On 02/01/2025 13:56, Sakthivel Velumani wrote:

I've also been told that you can build UHD apps directly on the X410--it
has all the tooling like Make, GCC, UHD and other important
 libraries already on the system image.

Hi Sakthivel,

As Marcus said you can run compilation of C++/UHD programs directly on
USRP X410 as it has all what is needed.

If you need a setup that is easier to integrate with DevOps you can
generate SDK for use on your PC using  infrastructure provided by yocto.
How to do that I've written in another thread ("KAS kirkstone build of
ni-titanium-rev5 on x410 with Vitis-AI ..."). As it was very long I'll
repeat the description here (for it to work you need to have docker and
git installed):

mkdir top_directory
chmod 777 top_directory
cd top_directory
git clone https://github.com/EttusResearch/meta-ettus.git
docker run --rm -it -w "$PWD" --mount
type=bind,source="$PWD",target="$PWD"
           --net=host
           --security-opt seccomp=unconfined
           ghcr.io/siemens/kas/kas:4.0
           bash -c "sudo apt update ; sudo apt install -y zip &&
./meta-ettus/contrib/kas_build_imgs_package.sh x4xx v4.7.0.0"

Then you can manually build sdk:

docker run --rm -it -w "$PWD" --mount
type=bind,source="$PWD",target="$PWD"
           --net=host
           --security-opt seccomp=unconfined
           ghcr.io/siemens/kas/kas:4.0
           bash

sudo apt update
sudo apt install -y zip
. oe-core/oe-init-build-env
bitbake gnuradio-image -cpopulate_sdk

This will produce an installer. You need to run it and install the SDK
somewhere.
Assuming it is installed in /opt/oecore-sdk before running
cross-compilation for USRP X410 with help of cmake you need first to
source the SDK's environment variables with following command:

source /opt/oecore-sdk/environment-setup-aarch64-oe-linux

Best Regards,
Piotr Krysik

W dniu 3.01.2025 o 00:27, Marcus D. Leech pisze: