Helping Medical Research With Your Computer's Spare Power

The Idea, and Where It Came From

When the pandemic hit, a lot of people wanted to do something and had no useful skill to offer. Folding proteins does not need volunteers in a lab. It needs compute, and compute is something ordinary people have sitting idle.

That is what distributed computing does: a research project splits a very large calculation into small pieces, thousands of machines each take a piece, and the results come back to the researchers. Your computer runs a small client in the background and never needs to know what the calculation is for.

The COVID work is the reason a lot of people signed up in 2020 and 2021. Most of them are still running it, and the research does not stop when the news cycle does.

A photorealistic shot of a laptop and a desktop tower on a home desk, both powered on, a small window with a task list visible on the laptop screen, evening lamp light, realistic photography, no legible text.

The Two Projects Worth Knowing

World Community Grid is a volunteer computing project originally funded by IBM and now run by the Krembil Research Institute at University Health Network in Toronto. Its focus has stayed on biomedical research: treatments for cancer, HIV/AIDS and neglected tropical diseases, plus work on low-cost water filtration and materials for solar energy capture.

COVID-19 research ran across its projects during the pandemic — the mpro and other viral studies were among the most heavily subscribed work on the grid at that point. If you want your cycles going to health research specifically, this is the project to pick.

Folding@home simulates protein folding and misfolding, which sits underneath Alzheimer's, Parkinson's, Huntington's and a long list of cancers. It is explicitly GPU-oriented, so if you have a discrete graphics card this is where that hardware does the most useful work per watt.

Both are free, both have been running for years, and neither requires anything from you beyond installing a client and leaving a machine on.

Setting It Up

Both projects run on BOINC, the open-source client from UC Berkeley. One client can attach to either project, or both.

  1. Download BOINC from the Berkeley site and install it.
  2. When it asks for a project, choose World Community Grid or Folding@home (or leave it and add projects later).
  3. Sign in or create an account for that project.
  4. In BOINC's computing preferences, set the CPU and GPU limits.

The last step is the one people skip. Defaults will happily use every core, which makes the machine loud and hot and is why a lot of installs get removed within a week. Set CPU to 60% and GPU to 50% — the work still gets done, just slower, and the research does not need your result by Friday.

Also set "compute only when idle" if you use the machine for work. Then it yields the moment you touch the keyboard and picks up again when you walk away.

A photorealistic close-up of a laptop screen showing a distributed-computing client with a list of research tasks and progress bars, a small globe logo beside each row, warm room light, realistic photography, no legible text.

A photorealistic shot of a desktop tower beside a monitor showing system temperatures and fan speeds, a dusty filter visible on the case intake, workshop lighting, realistic photography, no legible text.

What It Costs You

An idle desktop draws maybe 50-70 watts. Running CPU and GPU flat out can take that to 250-350 watts. On Australian electricity prices, an extra 200 watts around the clock is roughly 145 kWh a month — call it $40.

Throttled to 60% on a machine that is on anyway, the extra draw is far smaller, and the marginal cost is a few dollars.

Three practical cautions:

Heat and dust. A machine crunching continuously needs its fans cleaned more often than one that idles. I have lost a power supply to this after about two years of unattended running.

Laptops are a poor fit. Their thermal design assumes short bursts, not sustained load. Desktop or a small always-on server is the right hardware.

Phones are genuinely free. The Android BOINC client only computes while charging and above a battery threshold, which is exactly the right policy — a phone on a bedside charger overnight contributes for nothing.

Why It Is Still Worth Doing

Your single machine is a rounding error against the total compute of a research project. One laptop does not cure anything, and nobody is going to notice your contribution individually.

What changes is the aggregate. These projects are constrained by compute budget, and volunteer capacity is compute they do not have to buy. It is the same logic as a blood donation: marginal per person, not marginal in total.

And the setup cost is genuinely twenty minutes, once. After that it is a background client on a machine that was already running.

Don't trust the download button. Trust the hash.

If this was useful, More hands-on tech guides at TechGeek. No pressure — read it whenever it's useful.


Luke Keam
Spot a bug or want a custom utility built? I'd love to hear from you. Get in touch
Luke Keam | techgeek.biz