World Community Grid - View Thread - pseudo projects with a constant typical WU of corresponding real project

"I assume that it won't happen because it is not of interest for most of contributors..."

Not even that. WCG is not going to dedicate resources to that. Especially not during the transition.

Statement probably based on comparing runs of WU sets, including the noise of running not the same sets, which is what the suggestion would improve...
In theory HT could hamper over all performance: if L3 cache is much faster than RAM and full number (HT) of threads invalidates cache too fast and reduced number can run within L3 - no high latency RAM access... Needs to be tested - in Germany people say "Versuch macht kluch": experiments rule.

True, which is why ARP is an exception to the "HT means more work done".

Operating system matters for *some* projects. ...
Statement probably based on comparing runs including noise...

No. The Vina projects ran much faster under 64-bit Linux than under Windows. 50%-60% faster simply because of an OS change. This is true from beginning to end of these projects. (and this is also true for some other BOINC projects such as TN-Grid and Universe@home BHspin app).

I saw points as measurement of how valueable the result of a WU is - wrong?

I'm not a points expert. But every WU is valuable and every WU has a certain amount of work. I look at my desktop that runs MCM work units in about 2.5 hours and the points it asks for are basically the same as my laptop which runs MCM work units in 4.5 hours.

Faster RAM hasn't made any significant differences
Statement probably based on comparing runs including noise...

Just from observing people who, over the years, did some experiments. I also did such an experiment when OPN launched and found no benefit for that project. The progress speed reported by BOINC did not differ with RAM at 2600mhz (it's dual channel) or 1333mhz.
I run my RAM at 1333mhz to save a few (like 8) watts (so 667mhz each stick).

Most, if not all the "tricks" are already known. ARP is the only project where L3 cache and threads are something to account for. IIRC, you should run ARP on no more than half your threads.

I actually switched a laptop from Windows 10 to Linux Mint 20.2 less than an hour ago so I'll see if I can track the difference. It's an old AMD A8-4500M which ran 4 MCM workunits a few days ago. 3 took around 6.2 hours and another ran for 8.44. Small sample, I know.

MCM1_0181757_2377_0 Lenovo-G505S Pending Validation 2021-09-24 14:14:48 UTC 2021-10-02 14:14:48 UTC 2021-09-24 21:08:19 UTC 6.33 / 6.39 41.1 / 0
MCM1_0181755_8883_0 Lenovo-G505S Valid 2021-09-24 14:12:40 UTC 2021-10-02 14:12:40 UTC 2021-09-24 20:41:39 UTC 6.32 / 6.43 41.1 / 50.4
MCM1_0181750_7803_0 Lenovo-G505S Pending Validation 2021-09-24 14:12:40 UTC 2021-10-02 14:12:40 UTC 2021-09-24 20:43:47 UTC 6.35 / 6.46 41.3 / 0
MCM1_0181757_5898_1 Lenovo-G505S Pending Validation 2021-09-24 14:12:40 UTC 2021-10-02 14:12:40 UTC 2021-09-24 20:43:47 UTC 6.32 / 6.43 41.1 / 0

No major difference in runtime (yes, small sample) when compared to the Windows results I posted earlier. One thing of note are points. I believe Linux usually gets less points - I remember SCC also got less points under Linux vs Windows. I think I remember there was a 5-10% speedup under Windows with MCM The difference is probably still there but my sample wasn't large enough. And it won't be since W10 is already gone and that laptop doesn't really run much.

Anyway, having been a long time volunteer at WCG, I think there is zero chance of these pseudo projects happening.

However, it may be possible to delete the checkpoints files and logs for a work unit so that BOINC can restart it from sratch. This would give you the chance of running the same work unit under different conditions, at least some of the conditions you listed (like RAM, HT on/off, etc).I believe I read about this before but I can't be certain. I've never tried that.