Dear all,
After the second batch of simulations the result were similar: 99.5% of the simulations endend satisfactorily; 0.5% returned apparently normal but with invalid values. These results are stable enough to be able to perform a larger test. We actually need it to know if this small percentage of errors is due to something in the code or something in the clients. Currently the number of users returning invalid results is so low that it is hard for us to figure out what the difference might be.

So, we have decided to continue to the second phase with a bigger test with 45000 working units. They are already going up. We hope they don't last long on the server!

If everything goes well in this second phase, we will probably consider the beta version stable and we will start launching the projects that we have prepared.

Thank you in advance. We will keep you update of the results.

Best,
Jesús.

===============================================================

Estimados usuarios:
Tras la segunda tanda de simulaciones el resultado ha sido similar: el 99,5% de las simulaciones finalizan satisfactoriamente; el 0,5 % arrojó resultados aparentemente normales pero con valores inválidos. Estos resultados son lo suficientemente estables como para poder realizar una prueba más grande. De hecho, la necesitamos para saber si este pequeño porcentaje de errores se debe a algo en el código o a algo en los clientes. Actualmente, la cantidad de usuarios que devuelven resultados no válidos es tan baja que nos resulta difícil determinar cuál podría ser la diferencia.

Por tanto, hemos decidido continuar con la segunda fase con una prueba más grande con 45000 unidades de trabajo. Ya están subiendo. ¡Esperamos que no duren mucho en el servidor!

Si todo va bien en esta segunda fase, probablemente daremos por estable la versión beta y empezaremos a lanzar los proyectos que tenemos preparados.

Gracias de antemano. Los mantendremos informados de los resultados.

Un saludo,
Jesús.

---

Jesús Carro
Universidad San Jorge
@InSilicoHeart

Ya tengo en mi mac ! ojala y no truenen :)

Gracias & bon courage :)

Great news! Thank you!

I have some trouble with the work fetch. My Mac would like to get some work, but no work is delivered:

Sat 23 Apr 11:35:50 2022 | DENIS@home | Sending scheduler request: To fetch work.
Sat 23 Apr 11:35:50 2022 | DENIS@home | Requesting new tasks for CPU
Sat 23 Apr 11:35:51 2022 | DENIS@home | Scheduler request completed: got 0 new tasks
Sat 23 Apr 11:35:51 2022 | DENIS@home | No tasks sent
Sat 23 Apr 11:35:51 2022 | DENIS@home | Project requested delay of 7 seconds

The issue is since the release of the batch. There is work available in the server status, but my Mac does not get any work.

A reset project does not changed anything.

Got some work. - Test applications was not enabled. But, not shure other BOINC project show a error that for the selected applications no work is available...

Because there is no more work :)

It didn't last long ;)

Probably this is the cause. The tasks were distributed very quickly, and it is not always visible on the web page because the cache (it takes an hour or so to show that there is none).But according to the statistics the Macs are getting tasks.

At this moment, the application is in beta status, do you have accept beta tasks checked? If you don't have it, the server won't send you tasks.

---

Jesús Carro
Universidad San Jorge
@InSilicoHeart

Jesús Carro wrote:

... 0.5% returned apparently normal but with invalid values.
... el 0,5 % arrojó resultados aparentemente normales pero con valores inválidos.

Did the tasks the fail validation happen to have been stopped at some point during processing and continued from a checkpoint?

In this run, I've noticed I was a wingman (task #3) to a WU and one of the hosts (213477) had an invalid task. This client has had valid tasks that did not continue from a checkpoint while all of their validation inconclusive tasks (at this time) have had a start from a checkpoint. Might just be correlation, but if this is a common theme, it might suggest something may be off in the way checkpoints are written and/ or read.

I'm happy to say that all is going well on my Mac.

At l'Alliance Francophone we are quite happy to see Denis coming back and hopefully we soon have another biomed stable boinc project !

Yes, all the task that fail to be validated stopped and used the checkpoint. But many others that used the checkpoint were validated. So, yes probably it is caused by the checkpoint but in an specific situation that we need to find.

Thanks!

Jesús Carro wrote:

... 0.5% returned apparently normal but with invalid values.
... el 0,5 % arrojó resultados aparentemente normales pero con valores inválidos.

Did the tasks the fail validation happen to have been stopped at some point during processing and continued from a checkpoint?

In this run, I've noticed I was a wingman (task #3) to a WU and one of the hosts (213477) had an invalid task. This client has had valid tasks that did not continue from a checkpoint while all of their validation inconclusive tasks (at this time) have had a start from a checkpoint. Might just be correlation, but if this is a common theme, it might suggest something may be off in the way checkpoints are written and/ or read.

---

Jesús Carro
Universidad San Jorge
@InSilicoHeart

Yes, all the task that fail to be validated stopped and used the checkpoint. But many others that used the checkpoint were validated. So, yes probably it is caused by the checkpoint but in an specific situation that we need to find.

Thanks!

Jesús Carro wrote:

... 0.5% returned apparently normal but with invalid values.
... el 0,5 % arrojó resultados aparentemente normales pero con valores inválidos.

Did the tasks the fail validation happen to have been stopped at some point during processing and continued from a checkpoint?

In this run, I've noticed I was a wingman (task #3) to a WU and one of the hosts (213477) had an invalid task. This client has had valid tasks that did not continue from a checkpoint while all of their validation inconclusive tasks (at this time) have had a start from a checkpoint. Might just be correlation, but if this is a common theme, it might suggest something may be off in the way checkpoints are written and/ or read.

---

Jesús Carro
Universidad San Jorge
@InSilicoHeart

Any news??

The simulations ended and we are analyzed the results. If we detect that there is some space for improvement to achieve fewer errors, we will do so, but the numbers are quite good. We are preparing the information to launch the experiments we want to do. We will be a few days in the back preparing things. Probably tomorrow I will publish a news with some more information.

Many thanks.

Best,
Jesús.

---

Jesús Carro
Universidad San Jorge
@InSilicoHeart

Hi,
Thanks and good job !

Great!!

Looking forward to your update and Live steady tasks to work on.

Bill F