As topic starter mentioned, the inconsistency of credits awarded are hard to follow.
Having returned after some years, I am aware of the different systems of crediting used. (BOINC'CreditNew' also has its issues)
But within a given project I think it should be possible to have credits linked not only to amount of trickles, but also total CPUtime per WU, or have a fixed nr of trickles per WU. (See the diff. WAH2 vs HADCM)

20304628 10939599 1420079 26 Feb 2017, 8 Mar 2017 Completed 277,437.61 276,192.30 9,898.90 Weather At Home 2 (wah2)
Model runtime: 1yr,1mnth; 13 trickles
20304576 10907163 1420079 26 Feb 2017, 5 Mar 2017, Completed 340,442.84 339,084.90 4,355.42 UK Met Office HadCM3 short v8.29
Model runtime: 14yr; 1 trickle
20301285 10919421 1420079 26 Feb 2017, 13 Mar 2017, Completed 652,562.82 648,388.90 27,552.12 Weather At Home 2 (wah2) v8.24
Model runtime: 12yr,1mnth; 20 trickles

Will keep following the project, but as spectator.

---

I think it should be possible to have credits linked not only to amount of trickles, but also total CPUtime per WU,

This is how it works.

or have a fixed nr of trickles per WU

Trickles are per month or per year, and the length of models with the new programs can be long or short, without having to change the program.

Currently being run are these: New small batches of long runs --> with problems

If I get this wrong, I am sure someone will correct me but the number of trickles/WU is down to the scientists rather than the project people at Oxford Uni

I do know that the credits should be approximately the same/amount of computing power needed across model types but sometimes the initial estimates for this are out by a fair bit.

For me the important thing is that CPDN is a project that produces useful science and real results in an important field. (I won't say here which projects I personally think are frivolous or not useful) and I would still take part even if there were no credit system. For me, the main purpose of the credits is an extra check that things are working as they should.

If that is the case, please explain the difference between HADCM and WAH2 as put in my first post.
It is as Dave mentions, scores are not the purpose of running CPDN, but make it
easier to check (especially when running more projects) if everything is going well. Having to check per WU if trickles are sent (and received) etc. should not be necessary.
Perhaps things will improve after the ongoing IT transition is complete, untill so
I will let my models finish and graze on other pastures.

---

The first one in your list is actually a (wah2) (region independent) v8.12

The 3 program types in your list are all different, and all use different amounts of processor time/processor resources per unit.
These different amounts (per unit) are allocated different adjustment factors, which are stored in a look up table for each program type. When the credit script is run, one of the things that happens is that the number of trickles received for a task are multiplied by the adjustment factor for that model type.

So the amount of credit for a model depends on the number of trickles AND the program type.

*****************

Perhaps things will improve after the ongoing IT transition is complete

The IT work is purely infrastructure, and has nothing to do with the climate modelling.