Zero credit.. Why?
-
- Idle-Sammler
- Beiträge: 3
- Registriert: 22.08.2018 23:08
Zero credit.. Why?
I was wondering if there was a reason why I have lots of wu's that are granted zero credit. My wingman gets credit.
Re: Zero credit.. Why?
Please provide a link to such an workunit.
Re: Zero credit.. Why?
yoyo has made changes to the credit system which meant that some Linux hosts have been getting zero credits for work units that take a long time to crunch.
If you claimed more than 2000 credits for a wu over the Christmas holidays you got zero credit for the CPU time
I think the limit has been raised since then -- but there's no documentation...
This issue is caused by the BOINC client incorrectly calculating the measured integer speed for Linux hosts and yoyo using these incorrect values to calculate credit.
If you claimed more than 2000 credits for a wu over the Christmas holidays you got zero credit for the CPU time
I think the limit has been raised since then -- but there's no documentation...
This issue is caused by the BOINC client incorrectly calculating the measured integer speed for Linux hosts and yoyo using these incorrect values to calculate credit.
-
- Idle-Sammler
- Beiträge: 3
- Registriert: 22.08.2018 23:08
Re: Zero credit.. Why?
Your system claimed 5,386 credits and your wingman 223 credits. You claimed 24 times more than your wingman. The used CPU time of both results is nearly identical. This is not honoured by the system.
-
- Idle-Sammler
- Beiträge: 3
- Registriert: 22.08.2018 23:08
Re: Zero credit.. Why?
Okay shouldn't I also get the same as wingman? I have no idea why mine claimed so much.
Re: Zero credit.. Why?
This is still a problem. Example:
https://www.rechenkraft.net/yoyo/result ... =102289292
The run time for the task is correct, almost exactly a day in length. The claimed credit reasonable for a task that long. The wingman under claimed, in my opinion. In any case, awarding zero credits is clearly the wrong answer.
Edit: Until this gets resolved, the linux app(s) should be disabled. It is unfair to volunteers to get zero credits for valid results. It is just wasting their time and energy/money.
https://www.rechenkraft.net/yoyo/result ... =102289292
The run time for the task is correct, almost exactly a day in length. The claimed credit reasonable for a task that long. The wingman under claimed, in my opinion. In any case, awarding zero credits is clearly the wrong answer.
Edit: Until this gets resolved, the linux app(s) should be disabled. It is unfair to volunteers to get zero credits for valid results. It is just wasting their time and energy/money.
Team: SETI.USA
- joe carnivore
- Task-Killer
- Beiträge: 739
- Registriert: 04.05.2013 06:01
- Wohnort: Goslar, NI / Lower Saxony ,Germany
Re: Zero credit.. Why?
The wingman did not perform a benchmark.
This means the difference to your result is too big.
This is how I understood the previous answers.
This means the difference to your result is too big.
This is how I understood the previous answers.
Re: Zero credit.. Why?
I ran 44 tasks (44-core machine). All tasks that have validated so far resulted in zero credits.
Team: SETI.USA
Re: Zero credit.. Why?
Now i have also the problem with valid wus and zero credit.
https://www.rechenkraft.net/yoyo/result ... tid=508315
The limit for claimed credit makes no sense because the max granted credit is 1000.
The only thing we can do is:
disable Siever WUs if there are wus with long runtime,
or edit in the client_state.xml the fpops and iops to a lower value and disable the automatic benchmark in the cc_config.xml
Both are no good solutions.
https://www.rechenkraft.net/yoyo/result ... tid=508315
The limit for claimed credit makes no sense because the max granted credit is 1000.
The only thing we can do is:
disable Siever WUs if there are wus with long runtime,
or edit in the client_state.xml the fpops and iops to a lower value and disable the automatic benchmark in the cc_config.xml
Both are no good solutions.
Re: Zero credit.. Why?
..., or change to a different crediting system.
The fixed amount of credit that e.g. Einstein uses is working quite ok, I think.
Well, it's a bit on the low side for CPU, but it works.
And it definately doesn't just send out zero credit for work that has been positively validated.
Well, finding out what isn't ok and fix it is a fourth possibility.
Might be the best one.
Re: Zero credit.. Why?
So we are just going to leave the linux app active, even though it awards no credits? That is not nice to the volunteers.
Team: SETI.USA