Knightro2,
The last time knreed has posted about possible upgrades on the server side is there: Setting project priority

It's not exactly what you (and me too) would like to see but if your motivation is related to the characteristics of the WUs and not to the objectives of the projects there might be interesting improvements. But probably not really soon anyway.

Currently the only choices are
- either you select projects 1 and 2 and you will get a mixture of both
- or you select project 1 and monitor your Boinc client rather closely to see if you need to replace project 1 by project 2 in the related profile
(after unchecking the "if no work is available..." box in both cases).

Cheers. Jean.

It's always interesting to see the same subjects popping up in multiple threads, simultaneously. http://www.worldcommunitygrid.org/forums/wcg/printpost?post=212257

The everything else except CEP is quickly achieved by ticking all but HPF2 in the device profile and unticking the "if there is no work". I'm using all 4 profiles and assign to devices for best performance.

One of the things that we will eventually get to is the that for whatever reasons, some computers simply do not run certain projects correctly. Sometimes it is related to something that can be addressed via the science application and sometimes we cannot find a reason that it fails on that computer.

At the same time, BOINC has a mechanism that will limit the number of workunits sent to a computer if that computer returns nothing but error results. However, this applies across all applications.

I want to modify this so that it will look at the application level and if the client returns valid results for application a and b but not c, then as the number error or invalid results for application c increases, then it will be less and less likely that the computer will be assigned work from project c. This will never be a non-zero chance as the computer should periodically be given a workunit from that project to test if something has changed, but this would significantly limit the problems that are being discussed.

There are certainly many members who do look at the workunits being sent to them and how they are running and which ones are failing or are invalid and which ones run correctly. However, there are many more machines that are not monitored and an automated mechanism like this is highly desirable to maximize member contribution.

I saw this knowing that our list of things we need to get done is very very long (Forum Updates, Points Conversion, and a long list of other features) as well as new projects coming.

If you have more than one computer like retsof then sure you can play around with what computer does what units.

However I only have the one, which is a fairly stable machine only about 6 errors returned since I started crunching. I would like to do equal amount of work for all projects.

But I find when I select all project I alway seem to get loaded up with more of one or two projects and not as many as the others.

How does the scheduler decide who gets what and would it be possible to have a option to receive equal work for all projects.