well
project started recently
But the credits are much too low.
Perhaps for you this is not a priority at the moment.
review credit allocation
thank you very much

27 Apr 2025, 7:24:48 UTC 28 Apr 2025, 12:15:28 UTC Terminé et validé 18,415.89 15,118.56 5.18 Universal Docker app v1.08 (docker)
x86_64-pc-linux-gnu

27 Apr 2025, 7:24:48 UTC 28 Apr 2025, 14:48:35 UTC Terminé et validé 20,623.61 16,927.00 6.84 Universal Docker app v1.08 (docker)
x86_64-pc-linux-gnu

27 Apr 2025, 7:24:48 UTC 28 Apr 2025, 7:44:45 UTC Terminé et validé 1,799.42 1,465.00 0.40 Universal Docker app v1.08 (docker)
x86_64-pc-linux-gnu

27 Apr 2025, 7:24:47 UTC 28 Apr 2025, 0:41:54 UTC Terminé et validé 5,435.68 4,425.81 0.96 Universal Docker app v1.08 (docker)
x86_64-pc-linux-gnu

it's not enough

can you do anything to improve the credits?
it's a real catastrophe!!!

Please be patient.
We are working on making the functionality of BUDA applications stable.
Afterwards credit system will be reviewed and fixed if needed.

---

BOINC maintainer.
For any insight, check my BOINC Development Blog.

In reply to Vitalii Koshura's message of 2 May 2025:

Please be patient.
We are working on making the functionality of BUDA applications stable.
Afterwards credit system will be reviewed and fixed if needed.

ok
I'm patient
I wanted to draw your attention to the very surprising credits.....

I think we've fixed this problem going forward.
There were a couple of issues:

1) we were using BOINC's 'adaptive' credit system,
which keeps track of averages for hosts and app versions
and normalizes things so that different hosts get similar credit
for similar jobs.
This system assumes that jobs for a given app are roughly the same size.
That's not the case for the Boolean Chains app:
some jobs take 100,000 times longer than other jobs.
So we switched to a simpler system where credit is determined by
the runtime and the host's floating-point benchmark.

2) lots of hosts were reporting a 1.0 GFLOPS FP benchmark value,
meaning that the benchmark failed for some reason.
This exacerbated the problems of adaptive credit.
We need to figure out this problem.
In the meantime, we're using the average benchmark value (4.3 GFLOPS)
for these hosts.

In reply to David Anderson's message of 9 May 2025:

I think we've fixed this problem going forward.
There were a couple of issues:

1) we were using BOINC's 'adaptive' credit system,
which keeps track of averages for hosts and app versions
and normalizes things so that different hosts get similar credit
for similar jobs.
This system assumes that jobs for a given app are roughly the same size.
That's not the case for the Boolean Chains app:
some jobs take 100,000 times longer than other jobs.
So we switched to a simpler system where credit is determined by
the runtime and the host's floating-point benchmark.

2) lots of hosts were reporting a 1.0 GFLOPS FP benchmark value,
meaning that the benchmark failed for some reason.
This exacerbated the problems of adaptive credit.
We need to figure out this problem.
In the meantime, we're using the average benchmark value (4.3 GFLOPS)
for these hosts.

Thanks for the precision.