Deadlines are assigned by the individual projects, not by BOINC. ClimatePrediction, for example, has a deadline of next year. ProteinPredictor has a deadline of this week. Under the new 4.43 \"deadline\" scheme, PP will always get priority - and it did. After I upgraded BOINC to 4.43, PP was taking over all of my machines.

In the online preferences, I have assigned equal weight to all projects. I expect BOINC to honor MY decision, and automatically switch between projects every 2 hours, as I have instructed.

This new \"deadline mode\" is too smart for it\'s own good, IMO.

I have reverted back to 4.25 until BOINC drops \"deadline\" mode -or- better yet, gives me the option of using it or not.

--- bt

At least 4.43 does something to stop you missing deadlines. Yes you did a lot of work for predictor but after a while it would have built up a negative debt and stopped downloading predictor units in order to give CP a go. So it should eventually keep to YOUR resource share decision - it just takes a longer period to do so. I see it as 4.25 missing a crucial feature (having regard to deadlines) rather than 4.43 not working properly.

Deadlines are a problem for the project, not for me, and not for BOINC.

IMO, BOINC is solving the wrong problem, the wrong way, with "deadlines" and "debt".

I run five projects on each of my host machines, all with equal resource sharing. That's 4.8 hours each day for each project. PP estimates 2:15 for each WU. It can therefore process two WU's every day. I have BOINC set to "connect every 3 days". PP therefore queues up 6 WU's on my machine to process over the 3 next days.

PP would have to assign a really tight deadline for me to ever miss a deadline. Sorry, but NO project needs to have a result back in 3 days !! Some people only connect to the internet once a week. The expectations of the projects need to be "adjusted".

I like the "keep it simple" approach: BOINC can give each project a 2 hour window of CPU time on my machine. When the 2 hours are up, hand the machine over to the next project. What the project does with the 2 hours is it's business. Hopefully it makes efficient use of the time, and processes the WU's in FIFO order to minimize missed deadlines.

And just think ... deadlines only exist because the projects need to receive a quorum of results before it can hand out credits, which has nothing to do with "science".

--- bt

I looked up a table which told me Einstein units are 12 times bigger and have a deadline of 1 week. I figure that would give you a problem (in getting credits and providing useful work) if you joined that project. I could be wrong about that I only crunch CP and CP sulphur alpha.

I am not saying you cannot have it that simple way if you want. You can and have gone back to 4.25. You are entitled to your opinion but I thought it sensible to provide a different view so that others viewing this thread can look at what 4.43 does and make up their own minds. Without such an alternate view, people might think 4.43 has a serious problem.

Your point about wanting an option is IMHO a good one. Perhaps this should be put on the BOINC wish list - in which case the seti boards are the place to put it.

As well as complaining on the wrong project site, if you are after credits, then CPDN is NOT a good place to be at the moment, as the scheduler is miss-assigning models/credits.
And as you have several computers, you could easily have only one or two projects on each, so that there is no time related problem.

My "WISH LIST" item applied to the BOINC Manager, and since there is no BOINC forum, per se, I put my wish list item in all of the project forums (just to be sure).

I have 7 computers and all of them are doing a mix-and-match of the 5 projects.

CP crashed regularly on 2 of my machines and I have detached it from both of them.

The upgraded from 4.25 to 4.43 killed my half-completed CP workunit on another machine. The upgrade also caused a loss of communication on another machine so I had to detach and re-attach all of my projects. I'm back to 4.25 on all machines.

--- bt