Running TensorFlow on imx6sxsabresd

[deflight]

Hi i have imx6sxsabresd board.

My goal was to build with the tensorflow meta-ml package.

I built it with the following command according to the manual

EULA=1 DISTRO=fsl-imx-xwayland MACHINE=imx6sxsabresd source imx-setup-release.sh -b build-xwayland

mkdir ~/bin
curl https://storage.googleapis.com/git-repo-downloads/repo > ~/bin/repo
chmod a+x ~/bin/repo
PATH=${PATH}:~/bin

mkdir imx-yocto-bsp
cd imx-yocto-bsp

repo init -u https://source.codeaurora.org/external/imx/imx-manifest -b imx-linux-zeus -m imx-5.4.70-2.3.0.xml

repo sync

However, the meta-ml package was not included and built.

I looked up the relevant bb files.

armnn_git.bb

# We support i.MX8 only (for now)
COMPATIBLE_MACHINE = "(mx8)"

tensorflow-lite.bb

RDEPENDS_MX8 = ""
RDEPENDS_MX8_mx8 = "libnn-imx nn-imx"
RDEPENDS_MX8_mx8mm = ""
RDEPENDS_${PN} = " \
flatbuffers \
python3 \
python3-numpy \
${RDEPENDS_MX8} \

It was written as.

In the old technical consultation, I heard that TensorFlow is powered by a CPU on my board.

Is this not possible?

[weidong_sun]

I added and tested tensorflow-lite in Yocto for i.mx6sx. The meta-ml packages were needed during the compilation process, but the compilation failed.

[weidong_sun]

Yes, only i.mx8 serials processor support eIQ and machine learning. See below, please!

https://www.nxp.com/design/software/development-software/eiq-ml-development-environment:EIQ

Have a  nice day!

B.R,

weidong

[deflight]

However, I heard that eiq is not possible in imx6 solo, but tensorflow is possible in the latest version of bsp (imx-image-full). Doesn't this also work? So I built with bitbake imx-image-full.