I am trying to download WU for native Linux (nwchem, single thread) but am barely getting enough to keep machines busy. I have "test applications" and nwchem selected on my account (but not nwchem long)

What am I doing wrong please?

There are unsent tasks.

Is the priority of the project high enough?
Task storage settings?

DavidBAM has computers with more than 1 thousand processors , so the default settings of the project have to be adapted for his profile.
Link already provided by VENETO Boboviz : Job limits
But be carefull about the consequences on the server ability to manage the futher load.
DavidBAM should try to enter in the dance slowly...

DavidBAM has computers with more than 1 thousand processors , so the default settings of the project have to be adapted for his profile.

2400 processors in a single pc!! How it is possible??

DavidBAM has computers with more than 1 thousand processors , so the default settings of the project have to be adapted for his profile.

2400 processors in a single pc!! How it is possible??

52 computers.

DavidBAM has computers with more than 1 thousand processors , so the default settings of the project have to be adapted for his profile.

2400 processors in a single pc!! How it is possible??

52 computers.

I come up with over 65,000 total processors on 52 machines.
If each core got one WU, that would be more than 10% of all the available work.

2400 processors in a single pc!! How it is possible??

I would guess he is spoofing BOINC. I would ban it if I were running things.
Otherwise, if he can really do the work, I can leave.

DavidBAM seems to be involved in many Boinc projects.
Activities.

There are maybe other participants , with a similar computing powerfullness but with hidden computers , in order to avoid fear feeling or rejecting by other crunchers.

This is damotbe 's job to make each one work intelligently with other, defining restrictive laws which enable a real share of the work units available without non sense competition (job limits).

The goal is to do science and having the best diversified ecosystem to reach it.(Boinc spirit).

Just for information (Top500 supercomputers)

Great power implies great responsibility ...

What are the Boinc server limits ? What is the real admin endurance ? Only damotbe has an idea ...

I come up with over 65,000 total processors on 52 machines.

2400 core of Ryzen 3900 means 100 PHYSICAL cpu on a SINGLE machine.
Are you using blade servers? I don't know if exists server with Ryzen cpu (i know exists with Epyc cpu).
Suspect of cheating is strong

Instead of being hypnotysed by so big number of processors , i try to analyze the smaller computer [ID 1661] (only 1 processor):

I find impossible results :

1649532 1091399 27 Feb 2020, 7:32:39 UTC 27 Feb 2020, 8:55:09 UTC Terminé, en attente de validation 1,836.14 1,807.99 en attente NWChem v0.19 (t1)
x86_64-pc-linux-gnu
1649442 1091354 27 Feb 2020, 7:32:38 UTC 27 Feb 2020, 9:10:03 UTC Terminé, en attente de validation 3,089.40 3,050.44 en attente NWChem v0.19 (t1)
x86_64-pc-linux-gnu
1649443 1091355 27 Feb 2020, 7:32:38 UTC 27 Feb 2020, 9:22:32 UTC Terminé, en attente de validation 3,552.49 3,517.14 en attente NWChem v0.19 (t1)
x86_64-pc-linux-gnu
1649541 1091404 27 Feb 2020, 7:32:38 UTC 27 Feb 2020, 9:10:03 UTC Terminé et validé 2,886.62 2,848.72 200.00 NWChem v0.19 (t1)
x86_64-pc-linux-gnu

So if we calculate the maximum elapsing time : 9:22:32 - 7:32:38 = 6594s
How is it possible to completely run 4 work units (total running times : 1836.14 + 3089.40 + 3552.49 + 2886.62 = 11364.65) in 6594s.

The real value for this 1 processor is in fact 11364.65/6594= 1.72 processor approximated to 2 processors if the start of the work unit begins a little bit later than the reception moment.
So the number of processors displayed is false.
If it is wrong for this small computer , it may be wrong too for bigger hosts.

Thanks for your attempt to influence other crunchers, DavidBAM, we know now that we can't trust on these numbers.It was a good exercise...