Maybe they "forget" some hosts? In my case I have two running (Seti only) and only one is shown in
Boinc and Seti hoststats.
This in every progress of daily and incemental updates.

Willy
How is it that "SETI@home" hosts (697,315) exceed the number of "Combined BOINC" hosts (652,621)?
It doesn't seem right to me, unless I misunderstand the definition of "Combined BOINC".
Keith

PS: Having looked at the figures again, my guess is that 652,621 should read as 852,651. Then the figures would be proportionate with other statistics quoted. Keith

Willy, I appreciate that the way Combined SETI figures are calculated is correct (and appreciate that your site is the one which is doing this correctly), but I cannot understand how, under any system, one of the conntents (i.e. SETI@home HOSTS) can exceed the grand total (Combined BOINC HOSTS) which obviously must increase by a corresponding value every time any hosts are added. Am I not right to say that if there were no other hosts besides SETI@home, the Combined BOINC total would be the same value? If so, how can the addition of further hosts in other projects reduce this total? Puzzled, Keith
------------------------------------------------
Willy
I tried to make a reply to thank you for the explanation, but something seemed to go wrong. So I'm adding this message to my previous as a P.S.:-

So, it would seem that the SETI@home hosts Total is understated.
As it is the User figures that I'm monitoring, these would, I assume, be reliable?
Thank you again for the excellent reply and explanation.
Keith

P.P.S.:- The addition of the 156,923 with no CPID to the Total Combined BOINC Hosts would, though slightly overstated, be a much more accurate figure as very few of the early hosts would be duplicated.
I have amended the HOSTS which you quite rightly pointed out that I had incorrectly entered as USERS.
Keith

It's rather easy:

Old BOINC clients (prior to version 4 I think) did not have a host-CPID. These hosts are in the projects database (D without (= empty) CPID. Imported into BOINCstats, these still don't have a CPID (obviously).
The there are also some projects that have duplicate host-CPID's. This is an error in THEIR DB.

The CPID is the thing that matches the host across projects, and is (should be) a unique identifier. But, all hosts with an empty CPID are A] not unique and B] cannot be matched with another host because which of the thousands of hosts that also don't have a CPID is the match????

So, in the BOINC combined stats we only have the hosts that DO have a UNIQUE CPID. All hosts with either an empty CPID or a duplicate CPID are discarded.

A quick query on the SETI host DB (on BOINCstats) reveals 156923 hosts withoud CPID.

A duplicate search on the Einstein host DB (BOINCstats again) shows this (cpid, followed by number of entries, first 30 results only):

All these hosts are lost for the combined stats.

I hope this explains it better



Just saw you are talking about users, but I guess you mean the hosts as the BOINC combined stats have more users then SETI.

Sorry if this is in the bl**ding obvious category; have sympathy for a newbie. I have a common cross project id across projects, as I should, but the computer id for my 3 machines is different at seti to what it is at predictor. Could this be why, when I go to my host stats, the machine 'annie' shows up as ONLY working on Seti, which should not be the case; all 3 machines are set to, and actually do, work about 50/50 on each project. Total scores for annie are about 10000 if I add the scores from the individual sites, but not on boincstats. Look for user "annieb" (exact match) and my host stats. The machine 'myacer' should be about the same level as the machine 'annie'.

Comment?

Edit: -

Actually I just realised the computer names don't show in boincstats. 'myacer' is the 1700Meg machine, 'annie' is the AMD 2600+ machine and 'sebs_asus' is the AMD 4000+ machine. Note also that the AMD 2600+ is now showing the correct total credit (in the last couple of hours), but has some weird daily credit numbers.