How I found a way to help
I was watching a youtube video from Linus’ Tech Tips, when he mentioned the Foldin@Home project, and how lending your computing power to help them would help towards finding a vaccine for COVID-19. I went to their web site and explored setting it up, but I found it quite difficult to get going under Linux and the project seemed overwelmed by the new joiners and was struggling to find work. My computer spend the first few days doing nothing.
But it spurred me on to see if anyone else was doing similar, and I found the
Rosetta@home project. They were using software that was in my debian
boinc-client, with a user interface called
boinc-mgr), so I decided to try that.
Instant success. I was up an running in a few minutes, some jobs started downloading and my desktop machine was starting to work. Even when there were jobs using all 12 of my CPU threads (AMD Ryzen 5 3600) my machine was still responsive enough to work on most things.
I decided to see if there was the same software for my laptop, a Macbook Air.
Sure enough I found the software fore that and set that up too. However, running
it flatout on my laptop is not so good, as it made the fan run 24/7, and I don’t
fancy figuring out if you can replace a fan in an apple laptop in a few months
time, so I did have to turn it down a bit. Easy to, just tell
boinc to use less % cpu time and it automatically scales itself back.
Bringing the Raspberry PI into the equation.
A few days after I started all of this there were some people asking about whether it could run in a raspberry pi. Most people thought it would be pointless as the raspberry pi (or so they claimed) would not be powerful enough for anyone to develop software for it, but then Rosetta@home announced they had a dev team produce a 64bit arm version.
Now it just so happened that I had recently bought a 2GB Raspberry Pi4 to use for my Personal Audio Recorder Project and as a result by 4GB had no role to play. I decided to set it up to run these Rosetta@Home tasks.
My thanks go to Mark G James and his blog Marks Rpi Cluster and his post Do something useful with your Rpi4. He lays out all the steps set set it up as a client.
I set my raspberry pi up with the latest release of Raspian Buster Lite) and provided my self with the ability to seemlessly ssh into the machine (as the pi user). From there I updated the software and installed the
boinc-client package with the following commands.
apt update apt upgrade apt install boinc-client
I did one more thing before following the instructions in Mark’s blog
I editted the file
/etc/boinc-client/remote_hosts.cfg to add the ip address of
my desktop machine. This allows be to use the
boinc-mgr graphical control
program on my desktop to control the client on the headless pi server.
I used that to help me sign up to the Rosetta@home project and set the parameters for that machine I allow it to use 100% of the cpu time 100% of the time. I have cut down checkpoint of the work to every 600 seconds (from 60 seconds). It collects 0.1 days of work and up to and addition 0.5 days of work. The only other restriction I placed was to allow it to use 95% of the memory both when the computer is idle and when it is use.
I then followed Mark’s instructions (I won’t repost here because its better you visit his blog to find out).
My first pi to set up like this is doing exceptionally well. Its average daily credit about matches that of my macbook. With that success, I tried it on my 2GB raspberry pi 4. This has limited success as there is really not enough memory. I does get some work done but a lot less.
However I want to do my bit, so I have just purchased another 4GB Raspberry Pi and today set that running too.