[Python] Illegal instruction (core dumped) when running Streamlit app on Raspberry Pi 4

I’m trying to run a small Streamlit app on my Raspberry Pi 4. For testing, i made a small version with just core functionality:

# main.py
import pandas as pd
import streamlit as st

def main():
st.title("My Streamlit App")
st.write("Hello, world!")

df = pd.DataFrame({
'Column 1': [1, 2, 3],
'Column 2': ['A', 'B', 'C']
})
st.write("Here's a sample DataFrame:")
st.dataframe(df)

if __name__ == "__main__":
main()


On the raspberry, create a virtual environment, install Streamlit and run the app:

python3 -m venv .venv
source .venv/bin/activate
pip install streamlit
streamlit run main.py

Problem


When I try to run the app on the Pi 4, i get :

Illegal instruction (core dumped)


Sometimes even just doing:

python3 -c "import pandas"


inside the container causes the same crash. This happens both with my real app and with the minimal test app above.

What I tried

1. 
   Installing pandas via pip install pandas inside the container → still crashes with 132 Illegal instruction.
   
   
2. 
   Installing python3-pandas via apt-get works, but since i installed streamlit via pip, i get a module not found: pandas error.
   
   
3. 
   Adjusted a Dockerfile to install system dependencies (libatlas-base-dev, gfortran, build-essential, etc.) before pip install. Still crashes.
   
   
4. 
   Forced pip to use piwheels.org (pip config set global.extra-index-url https://www.piwheels.org/simple). Still crashes.
   
   
5. 
   Tried both approaches:
   
   • Debian packages (apt-get install python3-pandas) while leaving pandas out of requirements.txt. Still crashes.
   • Forcing source builds (pip install --no-binary=:all: pandas numpy) Installation failed.

Question


Has anyone successfully run a Streamlit app with pandas on a Raspberry Pi 4? What’s the correct way to install pandas (and numpy) so that it doesn’t pull in incompatible wheels with unsupported CPU instructions?

Any working example for Streamlit + pandas on Raspberry Pi 4 would be greatly appreciated.

If you need any more information, just reach out.

Thanks in advance!

Continue reading...