Bert Demo Error(s) about tt-buda-demos

[milank94]

Hi @danindiana

The version of pybuda you are running (v0.9.75) is not one of the stable released versions.

Could you please try on the latest stable release (v0.10.5) -- https://github.com/tenstorrent/tt-buda/releases/tag/v0.10.5.gs.240315

Before running it, check to see that the device is in a ok state by resetting the cores and trying the smoke test:

• tt-smi -r 0
• https://github.com/tenstorrent/tt-buda-demos/blob/main/first_5_steps/1_install_tt_buda.md#smoke-test

[danindiana]

Thanks for the link! Docker is up and running!

Howdy @milank94 so I attempted that pybuda build using 3.10.14 python on a fresh 20.04 build.

$tt-smi -r 0
Starting tensix reset on GS board at pci index 0
Lowering clks to safe value...
Beginning reset sequence...
Finishing reset sequence...
Returning clks to original values...
Finished tensix reset on GS board at pci index 0

Re-initializing boards after reset....
Detected Chips: 1
Detecting ARC: |
Detecting DRAM: |
[] ETH: |
(tsmi) smduck@calisota:/Downloads/tsmi/bin$ python3 --version
Python 3.10.14
(tsmi) smduck@calisota:/Downloads/tsmi/bin$

[milank94]

Closing issue for now due to inactivity. @danindiana please re-open if the issue persists.

[danindiana]

@milank94 Ok so some building is required in the docker container? Assuming the software is configured to utilize the card (non-trivial) and one runs the docker then what? I guess one can git clone the demos and then attempt to run those, etc. I'll try to watch George Hotz's video which I have been meaning to do but got busy.

root@9a800b7acc08:/home/tt-buda-demos/model_demos/nlp_demos/falcon# cat /etc/os-release
NAME="Ubuntu"
VERSION="20.04.6 LTS (Focal Fossa)"
ID=ubuntu
ID_LIKE=debian
PRETTY_NAME="Ubuntu 20.04.6 LTS"
VERSION_ID="20.04"
HOME_URL="https://www.ubuntu.com/"
SUPPORT_URL="https://help.ubuntu.com/"
BUG_REPORT_URL="https://bugs.launchpad.net/ubuntu/"
PRIVACY_POLICY_URL="https://www.ubuntu.com/legal/terms-and-policies/privacy-policy"
VERSION_CODENAME=focal
UBUNTU_CODENAME=focal
root@9a800b7acc08:/home/tt-buda-demos/model_demos/nlp_demos/falcon#

root@9a800b7acc08:/home/tt-buda-demos/model_demos/nlp_demos/falcon# python3 pytorch_falcon.py
2024-04-17 00:19:06.720 | INFO | Runtime - No silicon devices detected.
2024-04-17 00:19:06.721 | INFO | Runtime - No silicon devices detected.
Traceback (most recent call last):
File "pytorch_falcon.py", line 6, in
from nlp_demos.falcon.utils.model import Falcon
ModuleNotFoundError: No module named 'nlp_demos'

[danindiana]

I guess on the heels of that q here is another one:

mkdir pybud
smduck@calisota:/Downloads$ mv pybuda-v0.10.9.gs.240401-alpha-ubuntu-20-04-amd64-python3.8.zip /home/smduck/Downloads/pybud
smduck@calisota:/Downloads$ cd pybud/
smduck@calisota:/Downloads/pybud$ ls
pybuda-v0.10.9.gs.240401-alpha-ubuntu-20-04-amd64-python3.8.zip
smduck@calisota:/Downloads/pybud$ unzip pybuda-v0.10.9.gs.240401-alpha-ubuntu-20-04-amd64-python3.8.zip
Archive: pybuda-v0.10.9.gs.240401-alpha-ubuntu-20-04-amd64-python3.8.zip
inflating: manifest.yaml
inflating: pybuda-0.1.240401+dev.gs.c7ce1e7-cp38-cp38-linux_x86_64.whl
inflating: tvm-0.14.0+dev.tt.dfda3826c-cp38-cp38-linux_x86_64.whl
inflating: torchvision-0.16.0+fbb4cc5-cp38-cp38-linux_x86_64.whl
inflating: Dockerfile-release-ubuntu-20-04-amd64
smduck@calisota:/Downloads/pybud$ python3 -m venv env
smduck@calisota:/Downloads/pybud$ source env/bin/activate
(env) smduck@calisota:/Downloads/pybud$ pip install pybuda-0.1.240401+dev.gs.c7ce1e7-cp38-cp38-linux_x86_64.whl tvm-0.14.0+dev.tt.dfda3826c-cp38-cp38-linux_x86_64.whl
Processing ./pybuda-0.1.240401+dev.gs.c7ce1e7-cp38-cp38-linux_x86_64.whl
ERROR: Invalid requirement: 'torch@ https://download.pytorch.org/whl/cpu-cxx11-abi/torch-2.1.0%2Bcpu.cxx11.abi-cp38-cp38-linux_x86_64.whl; python_version == "3.8"'
(env) smduck@calisota:/Downloads/pybud$ python3 --version
Python 3.8.10
(env) smduck@calisota:/Downloads/pybud$ deactivate
smduck@calisota:/Downloads/pybud$ python3.8 -m venv env38
smduck@calisota:/Downloads/pybud$ source env38/bin/activate
(env38) smduck@calisota:/Downloads/pybud$ python3 --version
Python 3.8.10

[milank94]

@danindiana Sorry I'm not clear on what the last comment is referring to or referencing.

Just some pointers:

• If you are using Ubuntu 20.04, then the compatible version of Python will be 3.8
• If you want to run from Docker, then please use the steps highlighted here: https://github.com/tenstorrent/tt-buda-demos/blob/main/first_5_steps/1_install_tt_buda.md#docker-container-installation
• You will first need to install the driver (tt-kmd), update firmware (tt-firmware / tt-flash), and setup Huge pages
• This is the command to run the Docker container: https://github.com/tenstorrent/tt-buda-demos/blob/main/first_5_steps/1_install_tt_buda.md#step-4-run-the-container. Please note this step which ensure that the device gets mapped from the container: --device /dev/tenstorrent