Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
![]() |
World Community Grid Forums
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
No member browsing this thread |
Thread Status: Active Total posts in this thread: 83
|
![]() |
Author |
|
oliverstirling
Advanced Cruncher United Kingdom Joined: May 7, 2007 Post Count: 107 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
According to Knreed in this thread http://www.worldcommunitygrid.org/forums/wcg/...hread=25670&offset=20 there will be some more Beta's out for CEP shortly:
The next workunits coming are type 'C' workunits. We had to make some changes to the research application so we are currently in alpha testing. There will be a beta test soon and then we will be able to resume sending work out to everyone. ![]() |
||
|
nasher
Veteran Cruncher USA Joined: Dec 2, 2005 Post Count: 1423 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
i dont know if there is a way to FORCE beta's
----------------------------------------the best way i know to get them is to keep as few work units on your computer as posible therfore your turnaround time for jobs will be quick therfore when some showup your more likely to get them. another way(this may lead to idle time) is to choose a project with few work units going out and lots of people trying to get those.. this will mean you are likely to get no work from project (as i said may cause idle time) <<<not recomended but you asked for ideas personaly i just keep my cache of work units as low as posible and select the project(s) that i feel like running and every now and then i get beta work... i am up to 5 1/2 days completed beta... (when i had longer caches of work units i wasnt geting beta units) ![]() |
||
|
jasm580
Senior Cruncher USA Joined: Dec 20, 2007 Post Count: 157 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
is to choose a project with few work units going out and lots of people trying to get those This is what I do. There are a couple of projects that are w/o work right now. I have a days worth of work queued up and I am set to only get work from one of those projects. If a Beta starts today I will end up with some of them. I have to do some manual adjustments before my machines run out of work. -Jasm
-Jasm
|
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
You have made a couple of deadly mistakes in your suggested method. At this moment I'd suggest you read what I wrote and reply to that. Don't reply to what you think I wrote... For the rest, I realize you are a newbie. Ignorance is not a sin. Sorry to take so long to reply, but here is what you wrote: Now: how to force Beta Wu's: The moment you notice that there are Beta's available, change your buffer from 0.01 days to any higher number like 6, 8, 10 days. Your computer will automatically scoop up work for that longer period and so scoop up Beta's too. Right after the download, put your buffer back to 0.01 days and wait again... When the work you have is finished, do the same thing again. and this method is totally wrong. You will not get 6 days of Betas. What you will get is (possibly) one (maybe 2) Beta WU and the balance will be other non-Beta WU. When you realize your mistake, the natural tendency will be to abort the (approximately 20-30) non-Beta WU which then makes your system UNreliable and leads you into other problems. After 21 years of crunching, I think I no longer need to be called a newbie. |
||
|
gb009761
Master Cruncher Scotland Joined: Apr 6, 2005 Post Count: 2982 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Actually, probably a very good way of "increasing your chances of getting a Beta WU" at the current time, is to only have DDD-T selected and have the "if no work available, fetch from another project..." unselected.
----------------------------------------![]() Obviously, this will only work when there's a project that currently doesn't have any work to send out ![]() ![]() |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Actually, probably a very good way of "increasing your chances of getting a Beta WU" at the current time, is to only have DDD-T selected and have the "if no work available, fetch from another project..." unselected. ![]() Obviously, this will only work when there's a project that currently doesn't have any work to send out ![]() While that info may be too late for the round of testing currently running (since the WU's were already sent out), crunchers could assign one of the custom profiles (Work, Home, School) in Device Manager->Device Profiles to DDDT-only, uncheck both the opt-in to new and any available boxes, and any other desired options, then Save it. Don't forget to go to the Beta Testing page and check the box next to the new profile. Caveat: After creating your Device Profile(s) do not make any changes on the My Projects page ever again, because that page sets ALL of your profiles to the same settings you choose there. So, then... when you see Kevin announce a new beta test, switch a computer to use your DDDT-only profile, highlight WCG on the Projects tab and click Update, then bump its 'additional work buffer' setting up to 2.5 or 3 days, and wait for the beta WU's to be released. It should continue crunching the WU's from the task(s) you had selected previously. i.e. it won't really waste much crunching time sitting there doing nothing, but it will keep checking for DDDT WU's until it gets some beta WUs... leaving it on that profile all the time would create extra traffic (requests for work) at pseudo-random times (see next paragraph), so that would probably not be a 'good thing'. I haven't plotted it out or anything, but generally it looks to me like it checks, then waits a couple minutes, then about 5 minutes, then about 9 minutes, then about 11, then 15, then 20, then 30, then it might wait a few hours (possibly, THAT interval is tied to the 'Connect about every x.xxx days' time on the 'network usage' tab in Preferences; I certainly have never seen any OTHER connection interval that is actually tied to that setting) before it starts over checking every couple minutes and incrementing up - apparently the same algorithim it uses to retry uploads when the server is busy. |
||
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I haven't plotted it out or anything, but generally it looks to me like it checks, then waits a couple minutes, then about 5 minutes, then about 9 minutes, then about 11, then 15, then 20, then 30, then it might wait a few hours (possibly, THAT interval is tied to the 'Connect about every x.xxx days' time on the 'network usage' tab in Preferences; I certainly have never seen any OTHER connection interval that is actually tied to that setting) before it starts over checking every couple minutes and incrementing up - apparently the same algorithim it uses to retry uploads when the server is busy. Any "failed" connection like failing to connect to a server for upload, download or scheduler-request, but also asking for work but don't getting any, will give a random backoff between 1 minute and 4 hours. It starts with 1 minute and increases. But, then it comes to scheduler-requests, after 10 failed connections, the projects master-URL is re-downloaded, and this resets back to zero again, so next delay will be 1 minute again. (v6.6.xx and later clients has some additional timers that can influence things...) As for "Connect...", it's got multiple uses: 1: Keep atleast this much work cached, if project(s) can supply it and you're not being deferred for some reason. 2: Make sure you'll not get any work with shorter deadline than this. (overlooked in case of idle cpu) 3: If at all possible, make sure all work is finished "Connect..." + "Switch between applications..." before their deadline. (Used to decide then needs to switch to "High Priority"). 4: Any "Ready to report"-tasks is reported if less than "Connect..." until deadline. So, if beta-tasks has example a 3-day deadline, if "Connect..." > 3 days, you'll never get any beta-tasks at all, except if BOINC-client sits idle waiting on any sort of work. With a 3-day deadline, a setting of 2.5 days on the other hand would make sure any beta-tasks runs "high priority" nearly immediately after downloaded, and is reported (nearly) immediately. This means it won't be aborted due to quorum already met. (ok, if it is a 5-minute-task it can still be aborted...) ![]() "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." [Edit 1 times, last edit by Ingleside at Aug 14, 2009 12:49:08 AM] |
||
|
rembertw
Senior Cruncher Belgium Joined: Nov 21, 2005 Post Count: 275 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Sorry to take so long to reply Doesn't matter. The described system still is correct, as far as it's "huge" beta batches we talk about. With the huge batches it is possible to scoop up days worth of tasks. It was not a hold-by-hand walkthrough, but it could be deduced that it is interesting to increase the buffer in steps of 2 days until other tasks are coming in. At that moment, it could be deduced, it is best to stop increasing the buffer, and set the buffer back to minimum instead. When most of the beta's are crunched, repeat. With the more recent small batches, starting at about the time of my post you refer to, the dynamics are totally different. There the best way is to let Boinc contact WCG just as often as possible. In my opinion that is done by selecting 1 project only when there's rumour of a new beta batch starting up, being the project with the shortest tasks, and at the same time keeping the buffer as small as possible. When "feeling lucky" it might still help to increase the buffer somewhat temporarily as described above. There was also a recent batch, I seem to remember, that was not dumped in the send queue but trickled through over several days. In that case it was simply a matter of contacting as often as possible using a small buffer and small tasks so Boinc contacts often. Since beta's get priority, when they appear they will be cruched faster. If and when other tasks appear, they will be crunched after the beta's unless the deadline is getting near. By that time in general the beta batch is already processed. Hence there is no use in aborting these tasks if the start buffer was small to start with. The way described by gb009761 works too if you are willing to crunch "dry" or if you have other Boinc projects active. That system might still be improved by selecting no project at all, or only ask for new projects. Good luck with the beta's to all, regardless of the used "system"! There is not 1 system that's perfect for all situations, and I think the above sums up the possibilities pretty well. |
||
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
The way described by gb009761 works too if you are willing to crunch "dry" or if you have other Boinc projects active. That system might still be improved by selecting no project at all, or only ask for new projects. You can't choose to only run beta, you must select atleast one sub-project. ![]() "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
gb009761
Master Cruncher Scotland Joined: Apr 6, 2005 Post Count: 2982 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
You can't choose to only run beta, you must select atleast one sub-project. The way I've suggested, you wouldn't "technically" be running with just Beta - you'd also select DDD-T (which, just so happens, will have very few/if any WU's available, due to the recent completion of that particular project). ![]() |
||
|
|
![]() |