NVIDIA TLT Container for PPC64LE system

I am currently trying to install Nvidia TLT 2.0 on an IBM Power 9 PPC64LE system. I have install all of the prerequisites including the docker specific PPC64LE package and the PPC64LE specific NVIDIA Cuda container which is working properly. Every time I attempt to start the nvcr.io/nvidia/tlt-streamanalytics:v2.0_py3 container using:
"$docker run --runtime=nvidia -it nvcr.io/nvidia/tlt-streamanalytics:v2.0_py3 /bin/bash"
I get the error:
"standard_init_linux.go:211: exec user process caused "exec format error"".
If I run the container as detached, i.e.:
"$docker run --runtime=nvidia -itd nvcr.io/nvidia/tlt-streamanalytics:v2.0_py3 /bin/bash"
it immediately exits with the the most previous command being:
"install_ngc_cli.sh _"
This persists even if the container is restarted. I am wondering if there is a specific image:tag for tlt that I am supposed to use if I am using a ppc64le system but I cannot seem to find any documentation regarding this and there is nothing in the NGC catalog.

TLT is designed to run on x86 systems with a NVIDIA GPU such as a GPU-powered workstation or a DGX system or can be run in any cloud with a NVIDIA GPU.

For TLT 2.0, see Overview — Transfer Learning Toolkit 2.0 documentation

For error, please refer to Search results for 'exec format error #intelligent-video-analytics:transfer-learning-toolkit ' - NVIDIA Developer Forums

Does TLT 3.0 support the pcc64le CPU architectures or is that also strictly x86 systems?


The TLT is designed to run on x86 systems with an NVIDIA GPU (e.g., GPU-powered workstation, DGX system) or can be run in any cloud with an NVIDIA GPU. For inference, models can be deployed on any edge device such as an embedded Jetson platform or in a data center with GPUs like T4 or A100.