Our second new project write-up this week is a very different animal from the first. BitBOINC went public on 28 September 2026, declared itself fully operational the next day, and in roughly 48 hours had built up more raw throughput than projects that have been running for years.
📖 Full write-up on the BOINC Synergy wiki: boincsynergy.ca/wiki/BitBOINC
Let us get the obvious question out of the way
BitBOINC is not Bitcoin mining. It does not create new coins, it does not mine blocks, and it does not contribute to the operation or security of the Bitcoin network. The project is explicit about this, and so are we.
What it does is search for the private keys to a fixed set of publicly known addresses whose keys were deliberately placed inside specified numerical ranges as a public cryptographic challenge. In other words, it is a discrete-logarithm and brute-force search problem that happens to be denominated in bitcoin.
The Bitcoin puzzle addresses
The puzzle series is numbered sequentially: the private key for puzzle number n sits somewhere inside an n-bit range. The series runs up to puzzle 160, and many of the larger ones remain unsolved — which is unsurprising, since each step up the ladder doubles the search space.
The underlying mathematics is the standard relationship between a Bitcoin private key and its address. A candidate private key k becomes a public key through elliptic-curve multiplication, P = kG, where G is the generator point of the curve Bitcoin uses. That public key is then hashed down into an address, and a candidate only counts when the derived address matches the target exactly.
One detail crunchers tend to ask about: the project does not publish which subranges it has already swept. It publishes aggregate progress instead, on the reasoning that revealing searched locations would let unrelated searchers skip those regions for free.
Two applications, two strategies
- Potamos — a sequential search that tests private keys one at a time. Exhaustive, and the project describes it as practical mainly for smaller key ranges.
- Keraunos — an implementation of Pollard's kangaroo algorithm, a discrete-logarithm method. It can be used when the target's public key has already been exposed on the blockchain, which happens when the address has previously sent a transaction.
Each application works a single target at a time. When a target is solved, or when the project determines someone else has claimed the coins, the server advances to the next one in the queue. As of 29 September 2026, Keraunos was working puzzle 140 and Potamos was working puzzle 71.
Platform support is unusually broad
This is the part that will interest anyone with hardware sitting idle. BitBOINC supports:
- CPU: Windows, Linux, macOS and Android
- GPU: NVIDIA CUDA, OpenCL, and Metal on Apple Silicon
Android and Apple Silicon Metal support out of the gate is rare for a brand-new project. Actual availability is still subject to application and hardware support, but the spread is already wider than most established projects manage.
The project also reports that returned results are independently checked before being accepted and credited, and that credit has been switched to a native BOINC-style calculation based on the work your hardware actually performed.
Security, and what happens if a key is found
Given the subject matter, this deserves more detail than a typical project spotlight, so here is what BitBOINC states about its own handling:
- The application reads the work packet supplied by the project and writes its result inside the BOINC slot directory.
- It does not open its own network connections — all communication is handled by the BOINC client.
- If a private key is found, the result is encrypted immediately using RSA-4096 with OAEP padding.
- Only the encrypted result is stored on the server; the decryption key is kept offline by the project operator.
- There is no automated code path for transferring funds. The published procedure is for a person to retrieve and decrypt the result offline, independently verify that the key corresponds to the target address, and then manually transfer anything recovered.
💡 On the question everyone will ask next: BitBOINC describes a proposed distribution of any recovered funds across platform operation, science projects, nature and environmental protection, and participating volunteers. The volunteer share is described as a weighted random drawing among recent contributors to that particular puzzle — not an equal payout to every participant. Treat it as a stated intention rather than a guarantee, and crunch because the cryptography is interesting.
Early numbers
| Measure | Reported value |
|---|---|
| Performance | 29.312 TFLOPS |
| Active users | 39 |
| Total users | 39 |
| Active hosts | 113 |
| Total hosts | 115 |
Stats as of 30 September 2026 — two days after launch.
Thirty-nine users producing 29 TFLOPS tells you two things: the GPU applications are doing real work, and there is still an enormous amount of headroom for new volunteers.
Who runs it, and where to talk
BitBOINC is developed, authored and maintained by Alperen Yavuz and forms part of the Athena network. Support runs through the Libera.Chat IRC network on #bitboinc; the project recommends the Rhizome client for anyone who does not already have an IRC setup. The website also carries account registration, discussion forums, progress information, achievement badges and user levels.
How to join
- Install the BOINC client.
- Attach to https://bitboinc.athena.org.tr/ and create an account.
- Enable GPU computing in your BOINC preferences if you have a supported card.
- Set your resource share and let it run.
Links
- BOINC Synergy wiki article: https://boincsynergy.ca/wiki/BitBOINC
- Project site: https://bitboinc.athena.org.tr/
- About page: https://bitboinc.athena.org.tr/about.php
- IRC:
#bitboincon Libera.Chat
Update — and it’s a bigger one than I expected two days later.
BitBOINC is no longer just the Bitcoin puzzle hunt. It’s now running four applications, two of them brand new, and one of them isn’t cryptography at all:
Sphinx (live since 29 Sep) — same collaborative kangaroo search as Keraunos, but pointed at Certicom’s ECC Challenge instead of Bitcoin. No coins, no wallet, no address anywhere in it — just a 1997 elliptic-curve discrete-log problem that’s been standing for 29 years. Current target is ECCp-131 ($20,000 prize), with ECCp-163 and ECCp-191 queued and ECCp-239/359 planned once the wider-precision math is in. Credit where it’s due: the project says up front it has not confirmed those prizes are still honoured, and tells you to treat a solve as a cryptographic achievement rather than a payday. Refreshing bit of honesty.
Kampylos (announced 30 Sep, beta) — actual astrophysics. Exhaustive binary-lens gravitational microlensing modelling: a background star’s light bent by a foreground pair of masses, where the χ² surface is multimodal and most published analyses stop at the first solution they find rather than mapping the whole parameter space. It runs over archival OGLE/MOA/KMTNet photometry, is built on VBMicrolensing (the standard tool, not something homemade), and the full solution maps go to Zenodo with notable results to RNAAS. Source is public: github.com/alplix/kampylos
Fair warning on Kampylos though — it’s genuinely still a test app. Work units have gone out and everything is coming back invalid; the dev has confirmed it’s a test build whose core code isn’t configured yet and that returned results are feeding his debugging. The GPU build is being held back over an FP32 bug. That’s caused some friction on the forums, with one cruncher arguing a project sending out work should grant credit for correct results, and others pointing out that beta WUs are normal when you’re a one-person project five days old without the budget for a separate Ralph/Albert-style beta site. Worth knowing before you set a resource share.
Growth has been steep. The 39 users / 113 hosts / 29.3 TFLOPS from the post is already out of date — as of today it’s 57 users, 196 active hosts and ~40.8 TFLOPS, with 46 new machines in the last 24 hours. GPU split is 85 NVIDIA, 89 AMD/Intel, 4 Apple Metal.
Keraunos is still on puzzle #140 and Potamos still on #71, with 76 more queued and roughly 903 BTC sitting in the remaining targets.
One correction to my own post: the Free-DC delisting isn’t a stats glitch on their end — the project flagged it on 28 Sep and says it’s exporting hourly in standard BOINC format but hasn’t been able to reach anyone there. Still unresolved, so signature stats remain patchy.
Wiki write-up has been expanded to cover all of it.