Automating Chia Plotting on Ubuntu

Chia plotting is both CPU and IO intensive. You can run several plots in parallel until you hit a bottleneck. Plotting in parallel will increase the processing time, but also the throughput. The goal is to maximize the number of plots per day. There’s a sweet spot after which adding more processes will decrease the throughput. Therefore, it’s important not to exceed this limit.

The following bash script is a primitive mechanism to maintain the optimal number of plotting processes running. It starts a new processes after a previous completes:

#!/usr/bin/bash

MAX=$1

if [ -z "$MAX" ]
then
  echo MAX not set
  exit 1
fi

MEMORY=28000
BASE=/home/x/chia-blockchain
BIN=$BASE/venv/bin
TEMP1=/big1
TEMP2=/big1/s1
DESTINATION='/media/x/My Book Duo/chia-farms'
KEY=1111111111
THREADS=2

current=`ps -ef | grep chia | grep plots | grep create | wc -l`

if [ "$current" -ge "$MAX" ]; then
  exit
fi

memory_per_plot=$((MEMORY/MAX))

cd $BASE
. ./activate

$BIN/python $BIN/chia plots create -k32 -n1 -t$TEMP1 -2$TEMP2 -d"$DESTINATION" -b$memory_per_plot -u128 -r$THREADS -a$KEY

The command line argument is the maximum number of plot processes.

The script is meant to be scheduled in crontab:

25,55 * * * * /home/x/chia-blockchain/create_plot 4 >> /home/x/chia-blockchain/create_plot.log 2>&1

You’d need to customize the following variables:

MEMORY: total mebibytes allocated for plotting. The script calculates the memory per process to avoid swapping.
BASE: Chia software installation directory
TEMP1, TEMP2: temporary directories for plotting, optimally on NVMe SSDs. It requires ~300GB per plotting process.
DESTINATION: farming directory, usually on cheap storage
KEY: wallet key
THREADS: the number of threads per plotting process

You can also use this script to find out the optimal degree of parallelism for your system.

After scheduling the script and attaching enough cheap storage for farming you can go away on vacation and hopefully find some coins when you get back 🙂