I have no technical knowledge to answer the apparent discrepancies. I simply installed the agent on 3 computers (don't know why stats show 4). They are all part of my home network. When I open the agent, it uses a "high-end desktop system" as the base for comparion, even though only one desktop is in my device list. In each of the comparisons, I show up at 196 or better. These are my devices:

Alienware desktop--3.8 Mhz, 4G RAM, 100% throttle
Alienware laptop (desktop replacement)--3.6 Mhz, 3G RAM--80% throttle
Toshiba Tablet--2.7 Mhz--2G RAM--100% throttle

All operate 24/7, 365. Am I doing something wrong except to throw massive processing time into the project.

SNLCSJwillows-VM, you seem to be making a few assumptions. First of all, work units come in different sizes. Over the years, work units have got larger and smaller, as well as the day to day variation in each unit or batch. Members contributing to different projects can expect to end up with very different averages per work unit.

As for runtime, multi CPU devices can do twice or four times as much work in a single day. There are even a few 12 CPU devices running on the grid. Also, the BOINC agent awards credit on the day a work unit is validated, not when it is returned. All this makes it very hard to draw conclusions from a single datum.

WCG are aware of the weaknesses in the statistics display, but they are not errors.

No idea who is being compared with whom, but unless u know the detailed info of a members choice of projects, the type of computer(s), the change of computers (speed and cores e.g.), the agent, whether suffering a period of very long WU's hanging or cycled thru WU's in moments (which would register as a result under UD agent) and a few more, there is absolutely no comparing and loads of assumptions enter the equation.

For the exercise pulled all the participants together in this thread, no one doing 572 WU's a day in this set, but there are those that do literally hundreds..... just look up RCTCgrid (added for amplification), who just en mass switched from UD agent to BOiNC and look at the Points per hour...... they are all on the low side, considering that there are machines that genuinely generate 80+ per hour, per core. Considering that I e.g. added a BOiNC agent April 06, which runs a benchmark of 8 per hour (56 UD points) and running 24/7 at 90%+ CPU efficiency, the 41 UD points per hour, is actually on the low side.....because they are a historical build from day 1.

Discussion Participants comparison

As for the team switch observation, when a member moves from one team to another, or was team-less and joins one.... the points only count for a team the moment joining a team until quitting a team.

To sum it up, i will not assume anything in these individual stats matters, know the device count is just a historical count of registrations, know that Sid twice registered the same computer (he 'told' us), know that from experience, views of anyone's hourly points is an 'opinion'.

cheers

PS, i know there is a team comparison page like there is for members, but never used it, nor do i know who to compare in this thread..... unless on BOiNC, it's not easy to find out someone's team.

L'Incantevole Abitudine!