Posts by robert.mouris

1) Message boards : SZTAKI Desktop Grid : No work! (Message 8271)
Posted 1392 days ago by robert.mouris
It has become so hard to get new WUs and my cores are intermittently thirsty. Is it a glitch or a new policy to keep the queues low? Or is the project about to reach its goal? Looking at the second graph of the performance statistics is depressing.

2) Message boards : SZTAKI Desktop Grid : Initial replication is more than minimal quorum (Message 8180)
Posted 2185 days ago by robert.mouris
Good news! The BinSys work units created today are issued without a fourth replication.
3) Message boards : SZTAKI Desktop Grid : Initial replication is more than minimal quorum (Message 8161)
Posted 2224 days ago by robert.mouris
I just got 16 WUs. They were created yesterday and all of them have an initial replication of 4/3. I expected 2/2. A few tasks sent to other people crashed shortly after processing start and the system immediately created replacement WUs.
4) Message boards : SZTAKI Desktop Grid : New Logo design (Message 7398)
Posted 3362 days ago by robert.mouris
My solution is ****** ******* **** ** ******* ** *****. (Two * for SZ).

Why is it only in English? Shouldn\'t a Hungarian project have it also in Hungarian, especially as there is plenty of available space in the logo?
5) Message boards : SZTAKI Desktop Grid : Announcement: Join the test facility to help the community (Message 6675)
Posted 3745 days ago by robert.mouris
Ádám, same for me.

6) Message boards : SZTAKI Desktop Grid : WU still pending (Message 6528)
Posted 3778 days ago by robert.mouris
In rare occasions, the WUs are short, the 3 crunchers start immediately and the validator agrees. Then this project is like any project with a quorum higher than 1.

Usually, however, the whole process is much longer, due to 3 reasons:

  • The processing length of a WU is unpredictable at its start, and this means that Boinc is not very efficient in calculating the number of WUs that a cruncher should get, as Boinc considers the length of previously crunched WUs. In order to compensate for this, Sztaki has a 28 day limit. Sometimes, a WU needs only a few minutes to be processed, but if 1 cruncher doesn\'t send in the results, the 2 fellow crunchers must wait for 28 days until the server understands this.
  • This project doesn\'t require fast results in order to create new WUs, unlike projects like LHC or the (dying) Chess960 project which need fast results. A project gets the most of its limited resources in the long term if the number of initially created WUs is kept to the minimum, i.e. the quorum number.
  • This project has created many WUs at the same time (too many IMHO) and these fill the waiting queue on the server.

Conclusion: in the long run, this project uses the resources in the most efficient way, but it is not attractive to crunchers who want validated results and/or credits immediately.

7) Message boards : SZTAKI Desktop Grid : WU still pending (Message 6524)
Posted 3778 days ago by robert.mouris
The word \"success\" means that a cruncher has crunched the WU and sent back a result file without problem. But the validator is still not happy with the outcome and wants a further result. The last crucher sent back the result on 24 July 2007 20:51:18 and the server created immediately a new WU on 24 July 2007 20:52:31. This hasn\'t yet been sent out because there is a long queue on the server. Look for instance at this WU that was sent out a few minutes ago after having waited for 76 days.
8) Message boards : SZTAKI Desktop Grid : expired replacement result not being sent (Message 6454)
Posted 3796 days ago by robert.mouris
There are so many WUs in the queue that they have to wait for a long time. Look for instance at this one, which was just sent out now, 2 months after its creation.
9) Message boards : SZTAKI Desktop Grid : No Credit For A Clean Run? (Message 6410)
Posted 3806 days ago by robert.mouris
Not only the wasted time is an issue, the result database might hold
a bunch of wrong results now.

Assuming that the uninterrupted results are correct and the interrupted
ones are incorrect, each of those 0-credits uninterrupted ones most
likely stands for one invalid entry (from the validated interrupted ones)
that made it into the science database.

p.s.: ... plus those WUs where 3 interrupted ones validated fine without
beeing correct - and no one will ever notice it.

As it seems that nobody is bothered by supposedly wrong results in the database, neither the project administrators nor the scientists who base their theoretical research thereon, I suggest that the quorum be set to 1. The project will advance much faster than 3 times as fast, as it will also bring back those who don\'t want to wait many weeks or months before getting their credits.
10) Message boards : SZTAKI Desktop Grid : The Wish / To-Do List (Message 6405)
Posted 3808 days ago by robert.mouris
So do I with 4 Windows systems, but not with hostid=223069.


Next 10 posts

Home | My Account | Message Boards

Copyright © 2017 SZTAKI Desktop Grid