I experiment all the time and break things too. That's my doing ;-)

Seriously, think this is not something that can be expected by the small WCG crew to be maintained as to start documenting which versions have what problems with what Hardware/OS/User Apps, the variations of host setups is endless. If at all, since this is multi-grid impacting, works for WCG but not for Einsteen, such initiative belongs at Berkeley.

Somewhere there was a post that the techs are advancing in upgrading the BOINC server side and at that time will match the device profile features with 6.10.??. I say xx because I'm not convinced that 6.10.58 will be the one adopted after skinning. Works though quite well on both my W7 and Linux OSsed clients. The current target was October to have this completed and promote the recommendation for Windows to 6.10 (mac and linux already are, but will probably go from 6.10.17 to that higher point release too. No hard info on that).

I have a friend in town who is a major WCG cruncher. He simply will not budge beyond what is on the WCG page. This is the smart decision.

I am running all WCG projects and a bunch of other BOINC based projects on five machines. I am running 6.10.43, quite usable if GPU's are not involved, which is the case with WCG.

What 6.10.43 gets me is the BOINC screen saver, and I can configure it to show no individual project screen savers. But what does show up is the current projects. That way I can glance at the screen as I pass by a room with a machine and I can see what's running.

No problems with this scheme.

No, I don't do it !
I am still using 5.10.45 (March 2008, if I remember correctly).
At the beginning of my participation, I followed "boinc advices" regarding new Boinc versions. Soon, I considered WCG recommendation only. Finally, I staid by 5.10.45, probably until the next installation of a new operating system (version).
According to my real time process and automation experience, I do not change a running system .
Cheers,
Yves

WCG isn't showing this kind of information, but atleast going by information both WCG and SETI shows, like OS according to BoincStats, the differences doesn't seem to be so large. So my assumtion is the WCG-client-usage is more or less similar, and the info would be interesting. Obviously there'll be more WCG-users using the v6.2.28-client than for other users, but if let's say 50% of WCG-users is using this client, it still likely means 30-40% is probably using v6.10.xx...

Now, based on RAC, to get an indication how many is active running an OS, the setup is like this:

OS - WCG% - seti%
Win - 84.37 - 86.88
Linux - 10.57 - 6.16
Mac - 5.01 - 6.69
Other - 0.06 - 0.27

WCG being so much higher than SETI in Linux is likely due to more users than normal is currently using Linux to run CEP2. Having similar info for the pre-CEP2-launch would be interesting, to see if there really was a marked increase in Linux-usage then CEP2 was launched.

Looking more closely only on windows, the grouping is like this:

OS - WCG% - SETI%
XP - 47.56 - 37.29
win7 - 19.33 - 39.65
microsoft - 12.86 - 0.17
vista - 12.23 - 15.47
2003 - 2.66 - 2.41
2003R2 - 1.82 - 1.62
2008 - 1.44 - 1.18
2008R2 - 0.97 - 1.14
2000 - 0.55 - 0.60
srv-longhorn - 0.41 - 0.17
XP-64 - 0.15 - 0.07
win-longhorn - 0.025 - 0.069
win2003 - 0.013 - 0.14
win98 - 0.0054 - 0.011
millennium - 0.0020 - 0.0025
NT4 - 0.0010 - 0.0027
win95 - 0.000002 - 0.00008

XP being higher for WCG than for SETI indicates WCG-users is slower to upgrade, and this is possibly also the case for client-upgrades.

There's also another huge difference, the "Microsoft"-listing is much higher for WCG than for SETI. Now, you'll maybe not aware of this, but v6.2.xx doesn't correctly report the OS for Vista, Win7 and Win2008, so you know 12.86% of WCG-windows-users runs v6.2.xx, and has one of these OS's. At the flip-side, you also knows that an absolute minimum of 33.96% of windows-users does not run v6.2.xx, and the minimum is only in the very unlikely instance that 100% of the XP-users, win2003-users and earlier OS runs v6.2.xx.

So, a minimum of 10.85% of WCG's daily production comes from computers running v6.2.xx, and a minimum of 28.65% of WCG's daily production comes from computers not running v6.2.xx. Last but not least, 72.53% of daily WCG-production from computers running Vista/Win7/srv2008 does not come from v6.2.xx-clients.


Yes I do remember this, but my assumption was that appart for WCG having a larger group of v6.2.xx-users, the other client-distribution is more or less similar for WCG to SETI@home, so could atleast give an indication of how things are also with WCG, and would therefore be interesting to see.

While the SETI-numbers isn't directly comparable, would still guess around 50% of WCG's vista/win7/srv2008-production comes from v6.10.xx-clients, seeing 72.53% of this production is non-v6.2.xx. If the WCG-numbers is 40% or 60% isn't so important, "around 50%" is still an indication many users is running fairly up-to-date clients.

So, if I read you correctly, Ingleside, you are stacking assumptions above assumptions** but you have no precise stats on BOINC versions used inside WCG.
Don't hesitate to come back when you have some, and meanwhile do click on your wingmen's Result Logs from time to time: that will give you a flavor (not a statistical view, I agree) of which oldies are still crunching around.

** Your weakest assumption being, as always, that the profile of credit hunters on many of the other sites is the same as the profile of WCG members only interested with having their agent smoothly running in the background for trying to help useful causes.
Much different motivations and interests.