Thread 'Infrastructure update: work supply and Windows telemetry'

Message boards : News : Infrastructure update: work supply and Windows telemetry
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
benjamin
Volunteer moderator
Project administrator

Send message
Joined: 23 Sep 26
Posts: 242
Credit: 54,980,907
RAC: 1,491,573
Message 136 - Posted: 3 Oct 2026, 14:37:55 UTC - in response to Message 135.  

Thanks both, these are great examples. Part of this already exists, just not on the BOINC site: all finds are tracked on our community dashboard, https://eternity-control-plane-prod.eternity-cp.workers.dev/. In the Users tab you can see each participant's boards per score (463, 464, 465…), their total and their best board, and every board itself is in the public library at https://eternity-board-library.pages.dev.

Does that cover what you had in mind, or would you rather have the count shown directly on your BOINC profile, like NumberFields? Let me know what's missing and I'll look into it.
ID: 136 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
mmonnin

Send message
Joined: 27 Sep 26
Posts: 42
Credit: 141,621,123
RAC: 4,172,786
Message 137 - Posted: 3 Oct 2026, 15:28:05 UTC

Ah I thought I had seen some other dashboard but couldn't find it again.

The request to not display computers doesn't seem to apply to the dashboard though

BOINC option:
Should Eternity II show your computers on its web site?
ID: 137 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
pututu

Send message
Joined: 28 Sep 26
Posts: 17
Credit: 15,948,746
RAC: 194,165
Message 138 - Posted: 3 Oct 2026, 15:51:49 UTC

In the community dashboard, there are two groups: "BOINC" and "native".
Two questions:
1. Anyway to sort the findings by "BOINC" group only?
2. Are there any search node duplication work between test two groups?
ID: 138 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
benjamin
Volunteer moderator
Project administrator

Send message
Joined: 23 Sep 26
Posts: 242
Credit: 54,980,907
RAC: 1,491,573
Message 140 - Posted: 3 Oct 2026, 23:14:15 UTC - in response to Message 137.  

You're right, and thank you for pointing it out. The "show your computers" setting wasn't applied to the community dashboard. It is now: for everyone who opted out, machine names, CPU models and OS are no longer shown. Your machines appear as "mmonnin · BOINC pc1", "pc2"… with only their work totals, and your account totals in the Users tab are unchanged.
ID: 140 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
benjamin
Volunteer moderator
Project administrator

Send message
Joined: 23 Sep 26
Posts: 242
Credit: 54,980,907
RAC: 1,491,573
Message 141 - Posted: 3 Oct 2026, 23:22:34 UTC - in response to Message 138.  

Good questions, thanks!

Done: the PCs and Users tabs of the dashboard now have an All / BOINC / Native selector, so you can show and sort BOINC participants only. Your choice is remembered in your browser.
No duplication. The search space is split into slots, and each participant, BOINC or native, works in its own slot. The server never hands the same slot out twice, so the two groups never search the same nodes.
ID: 141 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
pututu

Send message
Joined: 28 Sep 26
Posts: 17
Credit: 15,948,746
RAC: 194,165
Message 142 - Posted: 4 Oct 2026, 4:04:29 UTC

Benjamin, looks good to me. Not sure about anyone else. Perhaps you might want to clarify the meaning of each column but here is what I think the meaning of each column. Maybe post an explanation of each column in the web page for reference.

PCS = number of PCs active within 24 hours?
Active Threads = number of threads used within 24 hours?
Validated Rate (5 min) = Moving average number of nodes checked in G (billions of nodes?) within the past 5 mins?
Stabilized Rate (60 min) = Moving average number of nodes checked in G within the past 60 mins
Nodes = User total nodes searched in T (trillions of nodes) since joining?
Time = Total active run time by user (not by PCs?) since joining?
Finds = number of edge matching found for each level (starting from 463 to 480, hopefully)
Best = the highest number of edge matching count found by user

Maybe also to put the non-standard boinc stat link somewhere, perhaps in the Boinc message boards under the "Project" Forum?
ID: 142 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
benjamin
Volunteer moderator
Project administrator

Send message
Joined: 23 Sep 26
Posts: 242
Credit: 54,980,907
RAC: 1,491,573
Message 155 - Posted: 5 Oct 2026, 8:04:29 UTC - in response to Message 142.  

Thanks pututu, good guesses, and a fair point that the page should explain them. Here is what each column means:

PCs: all the computers grouped under your account, not only the ones active in the last 24 hours.
Active threads: threads computing right now. It shows "—" when nothing is running.
Validated rate (5 min) and Stabilized rate (60 min): search speed in nodes per second (G = billions per second), averaged over the last 5 or 60 minutes.
Nodes: total validated nodes searched since you joined (T = trillions).
Time: total time during which at least one of your computers was working. It's wall-clock time, so two PCs running at the same moment count once, not twice.
Finds: how many boards you found at each score (463, 464, 465…).
Best: your highest score.
https://stats.eternityathome.org, and a link from the BOINC site is a good idea.
ID: 155 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
pututu

Send message
Joined: 28 Sep 26
Posts: 17
Credit: 15,948,746
RAC: 194,165
Message 160 - Posted: 5 Oct 2026, 18:29:53 UTC

Benjamin, thanks to you too!

After thinking and reading a few more articles via google, perhaps it is also best to show the estimated time/probability/etc on how long it will take to reach the end goal in the project main page where you have the overall running stat.

Taking a first shot at this. Let's talk in order of magnitude since we are dealing with humongous number here (much longer than the age of this earth in seconds).
→ Using optimized searching algorithm (I'm assuming that's yours?), there are about 10e+47 node steps to search? It's a reference point which I assume can be subjected to change if better a algorithm or faster hardware can be found in the future.
→ From live dashboard page, we are averaging about 200 billion nodes per second (native + boinc)
→ So far, we have searched 294,000 trillion nodes. That's insignificant when compare to 10e+47 nodes to check.

With all the information above, we can perhaps come up with a simple aggregate statistics to show the project progress. Something similar to cracking the RC5-72 72-bit encryption code https://stats.distributed.net/projects.php?project_id=8 where there are both boinc (aka moo wrapper project) and non-boinc contributors, similar to yours.

P.S. I know the number of days/years will be sheer mind boggling large and I think some contributors will either take this challenge head on or avoid it all together. Just to be honest here, not meant as a criticism ;)

From RC5-72 main page:
Aggregate Statistics

Total Blocks to Search: 1,099,511,627,776
Total Blocks Tested: 184,889,777,318
Overall Rate: 246 Blocks/sec
Total Keys to Search: 4,722,366,482,869,645,213,696
Total Keys Tested: 794,095,546,945,532,592,128
Overall Rate: 1,055,578,348,713 Keys/sec
Percent Complete: 16.816%
Time Working: 8,707 days

Current Information

56,330,658 Blocks were completed yesterday (0.005123% of the keyspace)(0.006159% of the remaining keyspace)
at a sustained rate of 652 Blocks/sec.

241,938,333,872,160,768 Keys were completed yesterday (0.005123% of the keyspace)(0.006159% of the remaining keyspace)
at a sustained rate of 2,800,212,197,594 Keys/sec.

The odds are 1 in 16,237 that we will wrap this thing
up in the next 24 hours. (This also means that we'll
hit 100% in 16,237 days at yesterday's rate.)
ID: 160 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
benjamin
Volunteer moderator
Project administrator

Send message
Joined: 23 Sep 26
Posts: 242
Credit: 54,980,907
RAC: 1,491,573
Message 161 - Posted: 6 Oct 2026, 7:07:47 UTC - in response to Message 160.  

Thanks pututu, I like the idea, and no offence taken at all!

You're right that the number of possible boards is astronomical. But we don't walk through them one by one. A lot of our work has gone into pruning the search tree early: cutting dead branches as soon as we can prove they can't lead anywhere useful. The algorithm builds on everything the Eternity II community has learned since 2007, plus new ideas we developed with the help of AI (ChatGPT and Claude), and the code itself is heavily optimised. So yes, the space is huge, but we eliminate most of it. Even then, it's very hard to put an honest number on how much is left. That's why a "percent complete" like RC5-72 would be misleading for us.

What we can point to is results. In the last few months the project went past the previous world record of 460 and reached 466, the best known score for a board that respects all five official clue pieces.

The current campaign gives every selected starting position ("seed") 10,000 search attempts, and that one has a real finish line. Once every seed has been tested, we'll pick the top seeds and focus the whole fleet on them to go after the record. We expect that to pay off.

Showing the progress of that campaign (seeds done, rate, expected end) next to the records is a good idea, and I've noted it for the dashboard.
ID: 161 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
pututu

Send message
Joined: 28 Sep 26
Posts: 17
Credit: 15,948,746
RAC: 194,165
Message 167 - Posted: 6 Oct 2026, 19:12:26 UTC
Last modified: 6 Oct 2026, 19:14:03 UTC

Good point. The percent complete will be meaningless when we are dealing with a super small number. Alternatively, just track the number of years instead.

Maybe start with a best guess for the upper and lower bound limits for the likelihood searched space and report a range of time needed to reach the goal, at least in your best judgement?
Taking a stab at this.

For example:
upper bound limit = 1e+47
lower bound limit = 1e+42
current run rate = 2.0e+8 nodes/second
in a year = (2.0e+8) * (3.2e+7) nodes searched per year. Let's round this up to 1e+16
estimated shortest time needed to find 480 matching edges = (1e+42)/(1e+16) = 1e+26 years
estimated longest time needed to find 480 matching edges = (1e+47)/(1e+16) = 1e+31 years

Over time, perhaps better guesstimate for the upper and lower bound may emerge.

P.S. I'm using writeup from wikipedia https://en.wikipedia.org/wiki/Eternity_II_puzzle and round down to 1e+47 as the upper bound limit and pulled something out from the thin air as the lower bound limit which may not reflect your optimized search algorithm.
According to the mathematical game enthusiast Brendan Owen, the Eternity II puzzle appears to have been designed to avoid the combinatorial flaws of the previous puzzle, with design parameters which appear to have been chosen to make the puzzle as difficult as possible to solve. In particular, unlike the original Eternity puzzle, there are likely only to be a very small number of possible solutions to the problem.[5] Owen estimates that a brute-force backtracking search might take around 2.0e+47 steps to complete.[13]
ID: 167 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
benjamin
Volunteer moderator
Project administrator

Send message
Joined: 23 Sep 26
Posts: 242
Credit: 54,980,907
RAC: 1,491,573
Message 168 - Posted: 6 Oct 2026, 21:02:19 UTC - in response to Message 167.  

Nice back-of-the-envelope work, pututu! One small correction: the project runs at about 200 billion nodes per second, so 2×10¹¹ rather than 2×10⁸. That's about 6×10¹⁸ nodes a year, which brings your range down to roughly 10²³–10²⁸ years. Still mind-boggling!

The honest caveat is that nobody knows how many complete 480/480 solutions exist, or where they sit in the search tree, so any "time to 480" is a guess built on a guess. That's why we measure progress in records: from 460 to 466 so far, all five official clues respected. And because our search prunes most of the tree early, the raw size of the space says little about how fast we climb.

Still, a "fun facts" line with your kind of estimate, clearly labelled as an order of magnitude, could be a nice touch on the dashboard. I'll keep it in mind next to the campaign progress idea.
ID: 168 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Previous · 1 · 2

Message boards : News : Infrastructure update: work supply and Windows telemetry
Powered by BOINC

© 2026 Eternity II Community