LHC@home

I used to be able to run many ATLAS WUs per computer but now I can only get 4 WUs to DL per computer. I have plenty of memory and can get a full complement of Theory WUs.
I run one CPU per WU. I've tried setting Preferences to ATLAS only and limit to 8 WUs or No Limit but the most I ever get is 4 WUs. Is it something I did or did LHC impose this limit?

---

What is your setting for Max # CPUs?

For Windows you need an app_config.xml.
Otherwhise only two or four tasks are avalaible.

What is your setting for Max # CPUs?

Like I said, one CPU per WU. Or as Preferences phrases it, "Max # of CPUs for this project" = 1.

For Windows you need an app_config.xml.
Otherwhise only two or four tasks are available.

The work server has no idea what's in your app_config.xml file. Or has that changed?
BTW, I only run Linux.

Like I said, one CPU per WU. Or as Preferences phrases it, "Max # of CPUs for this project" = 1.

That is your mistake. (The work server has no reasonable idea of anything you set, to answer your other question).

You set both Max # jobs and Max # CPUs to "No limit"

Then, you use the following app_config to run only one core per work unit (or any number you want):

<app_config>
<app>
  <name>ATLAS</name>
</app>
<app_version>
    <app_name>ATLAS</app_name>
    <plan_class>native_mt</plan_class>
    <avg_ncpus>1.0</avg_ncpus>
    <cmdline>--nthreads 1</cmdline> 
</app_version>
</app_config>

However, this trick works only for native ATLAS. I forget what you do for VirtualBox, but I think you just try at random until you find something more or less acceptable.

What is your setting for Max # CPUs?

Like I said, one CPU per WU. Or as Preferences phrases it, "Max # of CPUs for this project" = 1.

Max# of CPUs has a strange correlation with the max # of tasks. (bug serverside)
Increase it and reduce with app_config.xml the number of cpus to 1 again.
One negative point. BOINC will think you need more cores and claims more memory for that VM-task.