SETI@home
SETI@home (pronounced "SETI at home") is a volunteer distributed computing project run by the Berkeley SETI Research Center and hosted by the Space Sciences Laboratory at the University of California, Berkeley.[1] Its goal is to analyze radio telescope data in search of signals that could indicate extraterrestrial intelligence, making it one of many activities in the worldwide SETI effort. SETI@home ran as an active volunteer computing project from May 17, 1999, until March 31, 2020, when it entered an indefinite period of hibernation while the research team focused on analyzing accumulated data.[2]
SETI@home was the third large-scale use of volunteer computing over the Internet for research purposes, following the Great Internet Mersenne Prime Search (GIMPS), launched in 1996, and distributed.net, launched in 1997.[3] With over 5.2 million participants worldwide at its peak, it was the volunteer computing project with the most participants recorded to date,[1] and was recognized by the 2008 edition of the Guinness World Records as the largest computation in history.[4]
Background and origins
The concept for SETI@home emerged in 1995 when David Gedye, then a project manager at Starwave Corp., proposed using a virtual supercomputer composed of large numbers of Internet-connected computers to perform radio SETI analysis.[5] Prior to SETI@home, radio SETI projects relied on special-purpose supercomputers located at the telescope facility itself.[5] Gedye partnered with University of Washington astronomer Woody Sullivan, who suggested contacting Dan Werthimer, whose SERENDIP project was already conducting SETI observations at Arecibo, and with David P. Anderson, a specialist in distributed computing at UC Berkeley's Space Sciences Laboratory.[3]
The project was initially funded with just $100,000 from The Planetary Society and Paramount Pictures, and in its early years received donated server hardware from companies such as Sun Microsystems and Intel before transitioning to a purely donation-supported model.[3]
SETI@home was publicly launched on May 17, 1999. Within the first week, nearly 300,000 computers were already processing data from Arecibo. Within a few months, more than one million volunteers had signed up across 223 countries.[6]
Scientific goals
SETI@home was established with two primary goals:[5]
- To conduct useful scientific work in an observational analysis aimed at detecting intelligent life beyond Earth.
- To demonstrate the viability and practicality of the "volunteer computing" concept.
The second goal is considered to have succeeded fully: the BOINC platform, developed from technology pioneered by SETI@home, now supports dozens of computationally intensive scientific projects across many disciplines.[1] The first goal, as of 2026, has produced no conclusive evidence of extraterrestrial intelligence, though the project identified a number of scientifically interesting candidate signals for follow-up observation.[1]
How it worked
Data acquisition
SETI@home collected observational data "piggyback" or "passively" while Arecibo (and later the Green Bank Telescope) were being used for other scientific programs.[1] At Arecibo, data was sampled and written to high-density DLT cartridges at a rate of approximately one 35 GB tape per day. Because Arecibo lacked a broadband Internet connection, tapes were physically mailed to the SETI@home laboratory at UC Berkeley.[5] Once received, the data was divided both in the time domain and frequency domain into work units of approximately 107 seconds in duration, each roughly 0.35 MB in size, overlapping in time but not in frequency.[1]
Client software and screensaver
Volunteers installed a free client program on their computers. When the machine was otherwise idle, the program downloaded a work unit from the SETI@home server, processed it, and returned the results automatically upon the next Internet connection. The software also featured an optional screensaver that displayed a real-time visualization of the signal analysis in progress, showing spectrograms and signal-strength graphs.[7]
Signal detection algorithms
The client software searched for five categories of signals that distinguish genuine candidates from background noise:[1]
- Spikes in power spectra
- Gaussians: rises and falls in transmission power that may represent a telescope beam's main lobe passing over a radio source
- Triplets: three power spikes in a row
- Pulses: repeating signals possibly representing narrowband digital-style transmissions
- Autocorrelations: matching signal waveforms using autocorrelation
The core technique involved applying large numbers of discrete Fourier transforms (DFTs) at various chirp rates and durations, essentially equivalent to simultaneously tuning many narrow radio channels and looking for unexplained excess power. Formally, for a sampled time-series signal , the discrete Fourier transform at frequency bin is:
Each work unit was analyzed across hundreds of frequency sub-bands and drift rates to account for the Doppler frequency drift that would result from the relative motion between a transmitting planet and Earth.[1]
Work unit validation
To guard against fraudulent or erroneous results, every work unit was sent to multiple computers (a practice called "initial replication," typically set to a value of 2). Credit was only awarded once a minimum number of returned results agreed with one another (the "minimum quorum"). If disagreement occurred, additional copies of the work unit were distributed until a quorum was reached. The final credit granted to all machines returning the correct result was set to the lowest value claimed among them.[1]
From Classic to BOINC
The initial software platform, known as "SETI@home Classic," ran from May 17, 1999, to December 15, 2005. This program was capable only of running SETI@home tasks and required volunteers to manually download new software with each algorithm update.[1]
In 2005, the project transitioned to the BOINC platform, which had been developed by David P. Anderson with funding from the National Science Foundation.[8] BOINC allowed algorithm updates without requiring user intervention, enabled volunteers to contribute to multiple scientific projects simultaneously, and opened the door to new types of signal analysis.[9]
SETI@home Enhanced
On May 3, 2006, distribution of a new version called "SETI@home Enhanced" began. Taking advantage of increased desktop computing power since 1999, this version was approximately twice as sensitive to Gaussian signals and certain classes of pulsed signals as the original BOINC-based release. Application builds were also produced with processor-specific optimizations, especially for Intel instruction sets, allowing faster execution on compatible hardware.[1]
GPU acceleration
With assistance from NVIDIA, SETI@home developed a client application using the CUDA parallel computing platform. This GPU-accelerated version achieved speeds from 2 to 10 times faster than the CPU-only version, depending on hardware. For example, a GeForce GTX 280 was more than twice as fast as a high-end 3.2 GHz Intel Core i7 965 CPU running the same analysis.[10] GPU support via CUDA was formally incorporated into SETI@home in 2015.[1]
AstroPulse
AstroPulse was a companion application to SETI@home designed to search for short, broadband radio pulses in the same Arecibo data. Where SETI@home concentrated on narrowband continuous signals, AstroPulse used coherent dedispersion to search for brief but powerful bursts that could indicate rapidly rotating pulsars, evaporating primordial black holes, or previously unknown astrophysical phenomena, as well as another possible signature of extraterrestrial intelligence.[11] AstroPulse was one of the earliest test applications for BOINC. Beta testing of its final public release was completed in July 2008, and distribution of work units to qualifying machines began in
- ↑ 1.00 1.01 1.02 1.03 1.04 1.05 1.06 1.07 1.08 1.09 1.10 1.11 SETI@home. Wikipedia. Retrieved 2026-05-26}.
- ↑ (2020-03-07}).SETI@home hibernation announcement. SETI@home, University of California, Berkeley. Retrieved 2026-05-26}.
- ↑ 3.0 3.1 3.2 (2009}).SETI@home Celebrates 10th Anniversary. SpaceNews. Retrieved 2026-05-26}.
- ↑ Largest distributed computing project. Guinness World Records. Retrieved 2026-05-26}.
- ↑ 5.0 5.1 5.2 5.3 (November 2002}).SETI@home: An Experiment in Public-Resource Computing. Communications of the ACM. pp. 56–61. Retrieved 2026-05-26}.
- ↑ (2020-09-09}).How SETI@home accelerated alien hunting with an army of armchair astronomers. Digital Trends. Retrieved 2026-05-26}.
- ↑ SETI@home is on Pause. Universe Today. Retrieved 2026-05-26}.
- ↑ Anderson, David P..(2025}).SETI@home: Data Acquisition and Front-End Processing. The Astronomical Journal. Retrieved 2026-05-26}.
- ↑ SETI@home's transition to BOINC. SETI@home, UC Berkeley. Retrieved 2026-05-26}.
- ↑ NVIDIA CUDA Technology Dramatically Advances the Pace of Scientific Research. Retrieved 2026-05-26}.
- ↑ (2003-09-25}).New and Improved SETI@home. The Planetary Society. Retrieved 2026-05-26}.