You have selected to receive work from other applications if no work is available for the applications you selected

Whilst archae86 has more than adequately answered this, I'll add my bit just in case it helps further.

Yes, by choosing to run a more optimised app than the standard (stock) one. The advantage of the stock app is that staying up-to-date happens automatically but usually at a much later time than when the optimised apps become available. In the run just completed, the stock app (4.26) never did get updated to the best possible app (4.46) so people who chose manual intervention did contribute more efficiently for quite a long time.

In the newly started S5R4 run, the stock apps are the only ones available. They are essentially the best optimised apps from the previous run. There is no new further optimised app on the horizon yet but we could possibly reach that stage in a couple of months.

So why bother with the AP mechanism at all? The only reason is to crunch the remaining S5R3 tasks more efficiently than you would otherwise do. On the server status page (linked on the home page) you can actually see how many WUs for S5R3 are still out there. The current number is 61K. A WU consists of two successful tasks that need to be returned by two separate hosts. Probably most of the outstanding WUs only have 1 task outstanding but some will have two. So the total outstanding tasks is somewhere between 61K and 122K.

This number will steadily decrease but there will be a few bumps along the road. A fraction of the outstandings will not ever be returned so that more tasks will need to be generated as time progresses to replace them. My thinking is that if one of my machines receives tasks from this resend pool, I want to crunch it with the best optimised app available. The only way to do that is to choose manually.

I also see evidence that if my host is announcing to the scheduler that it can accept R3 tasks, there is a greater chance the scheduler will send it a suitable one. I've seen a few of my hosts receive multiple R3 tasks that the scheduler had been "sitting on" for a few hours, seemingly waiting for the appropriate host to come along and make a request. If the scheduler wasn't discriminating like this, the resends would always go out quickly to whatever host came along first. Quite a bit of bandwidth is saved if resends are sent to suitably equipped hosts.

You need to create your own app_info.xml by cutting and pasting exactly as described in the sticky thread. I have already modified the Windows example so that it will work for you. That file you create will tell BOINC about the apps to be used. You need to visit the beta test page (linked on the home page) and get the Windows 4.46 zip archive which contains all the apps that you need as listed in the sticky thread. You DO NOT need to download any tasks (work units) as they aren't special and BOINC will get what is on offer automatically. Please note that I haven't further mentioned the version 6.04 apps because they are already in your EAH project directory. Simply confirm to yourself that they are there and don't delete any other app files that might also be there.

Please reread the answers in this thread and the sticky thread that archae86 linked to. If this information doesn't make sense, you would be better just letting BOINC handle things automatically.

Many of us (probably all of us) are getting that message, whether or not we are using the anonymous platform mechanism (app_info.xml).

It is annoying but the good news is, despite the worrisome appearance, work is downloaded and gets done.

I assume, Alinator, you are referring to the comment from MarkJ

But, MarkJ, all four of your hosts show multiple new tasks sent on 12 August.

Most new task allocations in Einstein don't involve actual file downloads--but the transmission of a few bits telling your host to use files it already has downloaded slightly differently. This does not create the same logging imprint seen on SETI downloads, for example. Checking whether you "really" got fresh work is better done from the task list on the web site or in BOINCMgr than from the log.

These messages are indeed a bit distressing, and plenty of us have misconstrued their import to our hosts at first sight. I still think you are OK.