Create 3 vm using ./p2prc --CreateVM 0.0.0.0 --Ports=1 --GPU

then on each run 

sudo apt clean && sudo apt update && sudo apt -y install cmake ccache nano
sudo ln -s /usr/local/cuda-11.2 /usr/local/cuda
export PATH=${HOME}/.local/bin:$PATH && export PATH=/usr/local/cuda-11.2/bin${PATH:+:${PATH}} && export LD_LIBRARY_PATH=/usr/local/cuda-11.2/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
wget -c https://raw.githubusercontent.com/horovod/horovod/master/examples/tensorflow2/tensorflow2_synthetic_benchmark.py

pip install --no-cache-dir horovod[tensorflow]

# this will test if its locally install properly
horovodrun -np 1 python3 tensorflow2_synthetic_benchmark.py

then configure machine1 ssh/config file like
Host host1
  HostName 0.0.0.0
  Port 39751

Host host2
  HostName 0.0.0.0
  Port 34779

and ensure that the ssh can be connected and ssh keys are copied using ssh-copy-id command

then run 
horovodrun -np 2 -H host1:1,host2:1 python3 tensorflow2_synthetic_benchmark.py