Product Design, Manufacturing & Innovation Resources

Home » Online Tools » BOINC (Berkeley Open Infrastructure for Network Computing) Project & App

BOINC (Berkeley Open Infrastructure for Network Computing) Project & App

Calculation

Free
Download/Install required
No registration

https://boinc.berkeley.edu/

BOINC (Berkeley Open Infrastructure for Network Computing) is free, open-source software that lets people donate their computer’s spare processing power to scientific research. It was created at UC Berkeley’s Space Sciences Laboratory, originally to run SETI@home, and was released as a general platform in 2002.

How it works: You install the BOINC client and choose which research projects to support. When your computer is idle, or running at a limit you set, the client downloads small chunks of work from the project’s server, computes them, and uploads the results. The project then combines results from thousands or millions of volunteer machines. Redundancy (the same task sent to multiple volunteers) helps catch errors and cheating.

Our view: obviously it can be seen from both sides of the coin: as a calculation tool (for the scientists creating a calculation project for it), or as a sponsor (for users hosting the GOINC P2P tool to help on the calculations for the chosen projects … and also earn some Gridcoin as rewards).

What it’s used for: projects span astronomy (Einstein@Home searches for gravitational waves and pulsars), biology and medicine (Rosetta@home for protein structure, World Community Grid for health and sustainability research), climate modeling (climateprediction.net), and mathematics (projects searching for prime numbers and number-theoretic patterns).

Why it matters: volunteer computing gives research groups supercomputer-scale throughput at a tiny fraction of the cost. Many projects have contributed to published papers and real discoveries.

Practical points:

It runs on Windows, macOS, Linux, and Android.
You control CPU usage, when it runs (e.g., only when idle or plugged in), and which projects participate.
It does use electricity and generates some heat, so it’s worth setting sensible limits on laptops.
Participants earn “credits” for work done, which fuels friendly competition between individuals and teams.

The official site is boinc.berkeley.edu, where you can download the client and browse the list of projects.

Disciplines: Scientific computing, Distributed computing, Astronomy, Biomedical research, Volunteer computing.

Free alternative to AWS Batch, Azure Batch.

BOINC compared with AWS Batch and Azure Batch
BOINC	AWS Batch	Azure Batch
Compute resources: Volunteered computers running the BOINC client. One paper reports about 700,000 active devices and notes that the volunteer population has declined since the early years.	Compute resources: Compute capacity provisioned from EC2 instances or AWS Fargate, with jobs running on Amazon ECS, Amazon EKS or Fargate. EC2 Spot instances are an option.	Compute resources: A pool of managed virtual machines in which you choose the number and size of nodes and a Windows or Linux image. Azure Spot VMs can reduce cost.
Workload types: Independent work units sent to volunteer hosts. Applications can be native, wrapped executables, VirtualBox VM apps or Docker apps. Each job can run on two unrelated computers, and a third copy is created if the results disagree.	Workload types: Any job that can run as a Docker container, including array jobs and multi-node parallel jobs. Each job specifies vCPU and memory, and GPUs are available on P and G instance types.	Workload types: Embarrassingly parallel tasks such as Monte Carlo simulation and rendering, and tightly coupled MPI applications such as finite element analysis and fluid dynamics, using Microsoft MPI or Intel MPI.
Software licence and platforms: Open source under LGPL v3, and applications built on it do not have to be open source. Clients run on Windows, Mac, Linux and Android but not iOS.	Software licence and platforms: Managed cloud service, so there is no batch software or cluster to install. Jobs are submitted to queues and capacity is added or removed as needed.	Software licence and platforms: Managed cloud service reachable through Batch APIs, command-line scripts or the Azure portal. Scheduler or cluster software does not have to be installed or managed.
Cost model: The site describes BOINC as open source software offered for download, with no price stated. Volunteers earn credit for work done.	Cost model: No additional charge for the service itself. You pay for the underlying resources such as EC2 instances or Fargate.	Cost model: No extra charge for Batch. You pay for the underlying resources you consume such as virtual machines, storage and networking.
When it fits: Creating a project means staging input files on a public web server, writing an assimilator and signing application files with an offline private key.	When it fits: Jobs run as Docker containers, are submitted to queues and can be chained in pipelines or workflows with defined dependencies.	When it fits: Parallel or MPI jobs submitted by an application through an API. The documentation says to choose CycleCloud instead if you must keep an existing scheduler such as Slurm.

Privacy: we have no feedback of potential problem with this tool and author. But like many online services, you are here providing them with your own data. Depending of the criticity of your data, this is your responsibility to assess author's reputation and potential misusage, and to check and accept their terms and conditions, or not.