Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummers-lowrise
|
Message boards :
Number crunching :
How do I change priority on what work units are worked first?
Author |
Message |
|
Hi
I want to help out with Distributed processing with some of the projects. I set up foe SETI@home as well as PrimeGrid. It looks like I am loaded with many Weeks worth of PrimeGrid work units, before I will ever get to a SETI@home work unit?How do I change the job priority if what gets worked on my machine?
Thank you
Martin Glazer | |
|
|
Hi
I want to help out with Distributed processing with some of the projects. I set up foe SETI@home as well as PrimeGrid. It looks like I am loaded with many Weeks worth of PrimeGrid work units, before I will ever get to a SETI@home work unit?How do I change the job priority if what gets worked on my machine?
Thank you
Martin Glazer
It's not as simple as that but one way is to go into Your Account, on the web page of each project, then Preferences for this project, then set the Resource share for so that your preferred project has more resources than your next favorite project etc. The PROBLEMS with that is that it's based on the recent daily average credits, so you can't set it so Seti does 3 workunits and PG does 1 workunit all the time every time, Boinc doesn't work like that. So for instance if you set PG at 100% and Seti at 50% then over the next month that's the number of credits, roughly, that each project will get compared to the other project. The problems in my example, Seti and PG, is that each give credits based on wildly different systems, so it will always be a problem trying to average things out. So since PG gives more credits than Seti does for a given workunit you will end up doing TONS more Seti units than PG units to get anywhere even close to the credits PG gives.
In cases like this, Seti and any other project, my suggestion is to manually run Seti for August, then PG for September, then Seti for Oct etc, etc. ORRR get more pc's, like alot of us, and put each pc on it's own project!! Then as your interests change you can move them around, or even focus on one project and do ALOT of work for just it. | |
|
|
Another very crude means if you are working with a much shorter time scale (say day to day) is to suspend a project via the projects tab in the BOINC manager so that the other will run on its own and vice versa. | |
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13513 ID: 53948 Credit: 236,922,854 RAC: 3,199
                           
|
In my experience, BOINC does a terrible job of mixing projects. It seems to never do what you want it to do.
You can micro-manage which jobs it runs by suspending tasks or projects, but that gets tiring after a while. What I do is to only run one project at a time on a computer. "All of the above" isn't really a good choice with BOINC.
____________
My lucky number is 75898524288+1 | |
|
|
In order to assign a "fair share" to each of your projects, you might limit the number of wu's running concurrently per project.
This can be stipulated in a file app_config.xml in their respective project directories C:\ProgramData\Boinc\projects\www.primegrid.com, and I'm almost sure SETI has its own directory very near.
Contents like this:
<app_config>
<app>
<name>camb_boinc2docker</name>
<max_concurrent>1</max_concurrent>
</app>
</app_config>
Boinc2docker is another subproject [cosmology], substitute a name found here
https://www.primegrid.com/forum_thread.php?id=5650&nowrap=true#75618 for Primegrid.
With a 4-core PC, <max_concurrent>2</max_concurrent> would give an equal share to both Primegrid and Seti.
App_config file in both program directories; I don't know what name Seti should be given; save the files as ANSI, a while ago UTF-8 was read but not understood by BOINC.
After creating those files, let BOINC know it should abide by them. In BOINC manager: Options - Read config files - Click!
PS: I remember having seen this limitation (max_concurrent) somewhere being applied to projects, not subprojects as is done in app_config.xml. You'd have to consult the BOINC site https://boinc.berkeley.edu/wiki/Client_configuration to find out where and how it's done. | |
|
|
Another very crude means if you are working with a much shorter time scale (say day to day) is to suspend a project via the projects tab in the BOINC manager so that the other will run on its own and vice versa.
Something else which will help is the No New Tasks option in the projects tab of the BOINC Manager. With the first project having sufficient units to work with, set that project to No New Tasks and simultaneously activate Allow New Tasks for the second project so that new units (from the second project) will start to trickle in as units from the first project complete and which will queue up behind the existing units in progress.
Pay particular attention to my last statement above since projects (and subprojects) have deadlines which are not consistent across the board. If necessary, you activate Allow New Tasks for the second project as the cache is beginning to seriously dwindle down for the first project already in progress.
Of course, the reverse to all of the above is then applied once it comes time to flip from the second project back to the first project.
I hope all of the above makes sense. | |
|
|
thanks for the advice. I counted 69 PrimeGrid projects waiting, Its going to be 2 months before I ever get to work on seti@home. | |
|
JimB Honorary cruncher Send message
Joined: 4 Aug 11 Posts: 916 ID: 107307 Credit: 974,494,092 RAC: 70
                    
|
thanks for the advice. I counted 69 PrimeGrid projects waiting, Its going to be 2 months before I ever get to work on seti@home.
Please feel free to cancel any PrimeGrid work that your computer hasn't already started running. It'll be given out to someone else to run, quickly and automatically. | |
|
|
thanks for the advice. I counted 69 PrimeGrid projects waiting, Its going to be 2 months before I ever get to work on seti@home.
You might want to reduce your cache size too then, I have cable so I'm always connected so I have mine set at 0.5 for both the 'at least days of work' and the 'additional days of work' settings, that gives me about a 1 day cache of work on my pc. I also have a backup project for each pc set with a zero resource share that only gives me units when my main project is down or out of work. | |
|
|
JimB has already said what advice I would give. I can only speak for PG, but aborting an un-started task isn't a bad thing at all, if you can't finish it. Being unsure as to what to do with it is far worse; and there is always more work, so if you then decide that you do want to crunch that sub-project after all, then just get some fresh work once you've decided.
BOINC *IS* terrible, which of course is a real shame. As has been said; either micro-manage, or make sure you've decided what you want to crunch and configure all site and BOINC preferences so BOINC has no choice but to adopt a one-track mind until you tell it otherwise. Then it'll do the trick. | |
|
|
As long-time micro-manager (IIRC, I did at least three attempts for couple of month each with micro-management; I've quit entirely a year ago) I insist, that practice of micro-management is sure way to early death. Or, at least, alcoholism.
P.S. BTW, let's run micro-management challenge! That will be fun.
____________
I'm counting for science,
Points just make me sick. | |
|
|
As long-time micro-manager (IIRC, I did at least three attempts for couple of month each with micro-management; I've quit entirely a year ago) I insist, that practice of micro-management is sure way to early death. Or, at least, alcoholism.
P.S. BTW, let's run micro-management challenge! That will be fun.
The easy answer is just get more pc's, then put each one on it's own project and no micro managing needed!! You can get lots of good used pc's, Linux is free and getting Boinc to run on them is pretty failsafe nowadays. Heck I've even gotten pc's off the side of the road on trash days, people through away the strangest stuff sometimes!! The absolute first thing I do with them though is wipe the harddrive(s), no viruses coming my way nor me having ANY chance of see their leftover data, I don't even want to know!!! | |
|
Message boards :
Number crunching :
How do I change priority on what work units are worked first? |