Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Incapable of running btcli run --cuda on WSL ubuntu #897

Closed
urcades opened this issue Sep 2, 2022 · 2 comments · Fixed by #893
Closed

Incapable of running btcli run --cuda on WSL ubuntu #897

urcades opened this issue Sep 2, 2022 · 2 comments · Fixed by #893
Labels
bug Something isn't working

Comments

@urcades
Copy link

urcades commented Sep 2, 2022

Describe the bug
I have followed the bittensor installation guide verbatim on a Windows machine running WSL.

Everything was running perfectly, returning "working" results, up until I was asked to run btcli run --cuda to run bittensor on my GPU.

The initial errors I received had to do with cubit not being found as a dependency (patebin here), so I ended up installing it from source in the /.bittensor directory and pip installing it accordingly. I made some progress, but after running the btcli command with the cuda flag, I continued to receive stack traces telling me that a cubit.pyx file was not being found in my home directory. After copying the same file from the cubit repository I cloned earlier to my home directory, I've hit a wall and a stack trace I can't seem to route around.

This is where I've landed:

https://pastebin.com/BXwsQxs6

To Reproduce
Steps to reproduce the behavior:

  1. Follow the installation guide verbatim
  2. Run into errors after running btcli run --cuda

Screenshots
My first stack trace, saying cubit wasn't installed, even when I followed the installation steps successfully:

1

My second stack trace, after installing cubit from source:

2

My third, intractable stack trace, saying I need an integer:

3

Environment:
Windows 11 running WSL2:

DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=22.04
DISTRIB_CODENAME=jammy
DISTRIB_DESCRIPTION="Ubuntu 22.04.1 LTS"
  • Bittensor Version [e.g. 22]
    the latest release

Additional context
Add any other context about the problem here.

@urcades
Copy link
Author

urcades commented Sep 2, 2022

Wow, great timing:

After asking around in the discord for some help/insight, it turns out that running the btcli command with the following flags ends up working:

btcli run --cuda --cuda.dev_id 0

For whatever reason, omitting the --cuda.dev_id 0 flag ends up with all of the aforementioned errors.

Using the proper flags results in a seemingly working GPU registration process. It's currently solving now.

@camfairchild
Copy link
Collaborator

This is fixed when #893 is merged. Only happens for CUDA registration using btcli run without specifying the --cuda.dev_id flag

@camfairchild camfairchild added the bug Something isn't working label Sep 2, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants