I didn't exactly make multiple installs from the same device. I uninstalled an old and buggy client and installed an updated version. Seems to me the updated version of the client should have recognised the existing machine and not created a new machine ID, thereby duplicating machine entries.

Well, I had to uninstall the old version, because the version I downloaded by clicking on the "Download" link on the main page of WCG was (a) an out of date version, and (b) was the 32 bit version, and I have a 64 bit system with a high core count and it was maxing out at 32 cores. I wanted to utilise the full core count of the system and to do this I had to download the client from the Berkeley BOINC site instead, and uninstall the WCG version, which as I said was out of date anyway, as well as being 32 bit and installed in a different directory. So I see this whole mess as arising directly out of policies implemented by WCG and nothing that I did "wrong".

And I'm starting to get the feeling around here that somehow I'm not really all that welcome, it's all "my fault", despite being willing at my own cost and time to donate my free CPU cycles (probably more than a million points a day when other work is not being done).

Noetus,

You are most welcome and we are sorry that you have had troubles getting things running easily.

Just a couple of pieces of information to let you know why things are the way they are:

1) We link to a specific version of the BOINC agent because when we adopt a new version, we have run it through various forms of testing and security scans that go beyond the standard preparations done for a version of the BOINC agent. These are somewhat cumbersome processes on our side so we don't update our recommended version very often. However, we are due for an upgrade and we will be starting this work soon (after our next project launch). Your 56 core issue is definitely an example of the fact that we need to update.

2) There are a very large number of installation scenarios that the client has to handle correctly. One of them is when places like an institution like a University has a computer lab and they re-image their labs computers on a daily or weekly cycle. In these cases, the client is installed identically on all of the computer (with even the exact same contents of the data directory). Since tasks are assigned to a specific device, then if multiple computers think they are that device things then strange things can happen (including multiple computers running the same task). As a result, there are various checks made when a the software starts up and when it contacts the servers and if anything looks wrong, the system will treat the installation as a new device so that it gets its own work. These leads to multiple devices being listed in your profile, but it ensures the proper functioning of the system.

Hopefully you have made it past these startup issues and are able to participate without further issues. Welcome!

The device contact counter an-sich does not do that, speak the counter on the server is different than the one on the client, it will initiate an abort of the tasks on the client and assign new (if requested). Better to state it's a sum of a number of checked parameters. As members have posted, even recently, WCG's checks are passed if they move certain control files across hardware, even with different OSses, to include from Windows to Linux?]

But, the reason I came back here was this snippet "(after our next project launch)". The intel community has failed to start bouncing on this outlook statement ;o)))