I read that. But that means that many of my CPU tasks that where awarded 0.3 points have now been re-evaluated to 0.003. I think we all agree that that is ridiculous. A solid credit could be something like 60 points per hour.

Found and fixed the root cause. The experiment credit pipeline had a bug where experiment-mode results were being validated but skipped for credit assignment. This has been corrected — the AI validator now reviews and credits experiment results every 15 minutes based on compute time contributed and result quality.

The fix is currently being tested. If you have recently completed tasks showing 0 credit, they should be credited within the next few hours as the backlog clears. Going forward, new experiment results will receive credit within 15-30 minutes of completion.

You can check your credited results at https://axiom.heliex.net/results.php (new page, updates every 15 min).

Sorry for the delay on this — appreciate your patience.

We are back to no credits being awarded for tasks that complete successfully.

This is useless: https://axiom.heliex.net/results.php
Please restore our Computers and Tasks pages.

I don't understand a word of it but apparently this is what we're doing and why so many WUs are server aborted.
https://axiom.heliex.net/scientific_findings.php

You were right — I found a bug in the credit system. The AI reviewer was only looking at results named exp_* and completely missed GPU workunits (named gpu_wu_*). On top of that, when there were more results than the reviewer could process in one session, leftovers were never getting picked up.

Both issues are now fixed:
- All 7,000+ backlogged results have been credited (yours included)
- The reviewer now covers all workunit types
- Credit is distributed in batch so nothing gets skipped

You can check your credit breakdown per review cycle here: https://axiom.heliex.net/my_credit.php?userid=6

Sorry for the delay — appreciate the report.

Good feedback on both points.

results.php — The standard BOINC task list is back and working. I've also added a new credit dashboard that shows your credit grouped by AI review cycle — check it from your account page (Home → Tasks → Credit Summary) or directly at https://axiom.heliex.net/my_credit.php. Click any cycle to expand and see individual tasks.

Server aborts — Those happen for two reasons: (1) an experiment has collected enough data and retires, so remaining workunits are no longer needed, or (2) an experiment has bugs causing high crash rates. Experiments typically run for many cycles before retiring, so the vast majority of tasks complete and get credited normally. Only unsent or in-progress tasks at the time of retirement are cancelled.

scientific_findings.php — That page shows what the project is actually discovering from the experiments volunteers run. Each finding links back to the experiments and hosts that contributed.

Credit runs on a sliding 2-day window — the validator cycles every ~1.5 hours and credits all completed results from the past 48 hours in each pass. There was a gap between cycles while I was updating the pipeline, which is why you saw 0 credit temporarily.

Your completed tasks will be credited in the next cycle. The system distributes a fixed credit budget across all results returned in each window, so credit per task scales naturally with how much work you contribute — more results in a window means more of the budget goes to you.

If you still see 0 credit after a few hours, let me know.

In reply to PyHelix's message of 8 Mar 2026:

Credit runs on a sliding 2-day window — the validator cycles every ~1.5 hours and credits all completed results from the past 48 hours in each pass. There was a gap between cycles while I was updating the pipeline, which is why you saw 0 credit temporarily.

Your completed tasks will be credited in the next cycle. The system distributes a fixed credit budget across all results returned in each window, so credit per task scales naturally with how much work you contribute — more results in a window means more of the budget goes to you.

If you still see 0 credit after a few hours, let me know.

There is still a big problem with the credit system; it's not working at all.

Update on the zero-credit issue:

The credit system is working — results that show "0 credit" are simply waiting for the next AI review cycle, which runs every ~40 minutes. This is by design: our validator marks results as complete immediately, but credit is awarded in batches when the AI reviewer runs.

I've just updated the website so the status is clearer:
- Results waiting for credit now show "Completed, credit pending" instead of "Completed and validated" with 0 credit
- The Outcome column shows "Success (credit pending)" instead of just "Success"
- Hovering over any status now shows a tooltip explaining what it means

If you see "credit pending" — your result is valid and credit will be awarded on the next cycle. If you still see 0 credit after 2 hours, let me know.

You can also check your credit dashboard which shows pending vs credited tasks grouped by review cycle.

mmonnin — I can see you had a big fleet contributing throughout February (multiple 7950X, 9950X, 5950X machines). Unfortunately your old results were purged from the database during the credit system changes, so I can't recover the exact number.

Two questions:
1. Do you remember roughly how much total credit you had before the rescale? I'd like to restore it.
2. Are your hosts still running Axiom? I see your last contact was March 7 — if you're still crunching and not receiving work or credit, let me know and I'll look into it.

I want to make sure longtime contributors like you aren't getting shortchanged.

Before today I had over 60,000 credit. Now all I have is todays credit, the rest has been wiped out.

WTBroughton — sorry about that. The credit wipe happened during a system migration earlier this week. I have restored your 60,000 credit — your profile should now show ~62,285 (the restored amount plus today earnings).

Please let me know if the number looks right. Going forward, credit is awarded every ~40 minutes and will not be affected by system changes.

Thanks for the quick response. It puts me back roughly where I was.

Glad it looks right! For everyone in this thread — I have now implemented daily automated backups of all volunteer data (accounts, credit, teams, hosts, forum posts). So even if something goes sideways again, I can restore everything quickly. No more credit wipes.

I checked in this morning to restart this project.
I then noticed that i started today with 0 credit as well. Last time i checked is was 1.2 m.
And a lots of 0 credit jobs. That was why i took a break.

Not good if that happened to everyone ...

Restored your credit to 277,000. The 100x scale-down was applied to all users equally when I was reworking the validation system. Your relative position on the leaderboard is the same. It should never have hit zero, that was my mistake. I have since set up daily automated backups of all credit data so if anything goes wrong again I can restore quickly. Won't happen again.