Accessing Lambda Stack on Tensor Book

post by Emily55 on Apr 11, 2021

Hi Everyone,

I’ve just received my tensor book and was curious, how do I access Lambda Stack, where do I find the code?

Thanks!

Emily

post by fkx on Apr 15, 2021

New user here just like you, got my tensor book last week.

I’ve been using Nvidia Jetson systems with ubuntu 18.04 so I had some familiarity with ubuntu and a number of apps (VSCode, Python 3, Julia, etc).

Initially I was a little puzzled about where this “lambda stack thing” is.

I came to see it was just a set of consistent installs of many drivers, CUDA versions, and other packages that were just “there”.

The code is in the github though.

To see what you have, open a terminal and type the following:

sudo apt-cache depends lambda-stack-cuda

You’ll see a list of things that you can just start using.

Example:

~$ python3 -c "import torch; print(torch.cuda.device_count())"

will detect your GPU count and print it (1 most likely).
if it works then python3 and torch are installed Ok and working.

However, the following does not work (yet):

~$ sudo docker run --gpus 2 lambda-stack:20.04 python3 -c "import torch; print(torch.cuda.device_count())"

Gives the error:

Unable to find image ‘lambda-stack:20.04’ locally docker: Error response from daemon: pull access denied for lambda-stack, repository does not exist or may require ‘docker login’: denied: requested access to the resource is denied.

Implying something about the docker setup in the lambda stack.

So that’s what I’m researching.

I have Visual Studio Code installed and running but I suspect that somewhere along the way there will be a clash due to it not having been part of the “lambda stack.”
That’s OK, trying and failing until it works is a great learning experience if you can develop the tolerance for it.

I think it will be OK to add your experiences as you go along. That’s what I’ll be doing until somebody stops me for cluttering up their forum.