Help with best configuration for private cloud VMs

[dognodding]

Hi!

I have F@H running on some home PCs without any issues.

My employer has a private cloud environment which is currently idling and which we would like to "donate" for F@H for a couple of months, in particular to support COVID-19 research, until we are ready to bring it into service with our own applications. We have 40 12-core servers with a KVM-based OpenStack on top. As this system is of a different magnitude to my home setup, I have a couple of questions.

My first question relates to the most efficient setup for F@H on big machines: Is there an upper limit to the number of (v)CPUs that FAHClient can use? To get the most out of the available resources and reduce the effort to set up, we would prefer to run a smaller number of VMs on each physical machine, each with a higher number of virtual cores, but not if FAHClient does not make the most of this setup.

My second question relates to an optimal configuration: With a high number of (v)CPUs, wiill FAHClient automatically utilise all of them, or do we need to set the option client-type = bigadv to make use of them?

Any other configuration tips would be welcome!

Thanks in advance.

[bruce]

Machines with 12 (v)CPU threads are actually quite common in home machines today and it happens to be a really good number for FAHCore_a7. most of the questions we get about "big machines" are upwards of 32 threads.

Treat them like individual home machines. Unfortunately each machine downloads a unique ID the first time it connects, so if you plan to clone the install 40 times, do-so off-line. Don't let the master connect/start ... or install the software 40 times.

[dognodding]

Great, thanks!