The zPDT Guide and Reference redbook states that emily is the Linux process representing the CP.  I also see that there is only 1 emily process for each standalone or sysplex instance.  (We have 1 sysplex of two lpars and 4 standalone instances of zOS running on our server.)  I have also verified via ps aux and top that the emily processes are by far the largest consumers of cpu.

I understand the advantage of having 2 logical processors defined for a zOS system, but is there any advantage of defining more that 2?  Is emily single threaded or can it use multiple cpus?  Does emily use a prioritizing algorithm or just a round robin for dispatching its emulation?

Each emulated CPU is a thread within the emily process.  Normal linux dispatching takes care of the time slices and processor placement. Linux is smart enough to pin threads to real processors for performance reasons.  But the scalability of zPDT is hindered by emily having to maintain System z memory consistency and by emulating System z broadcasting functions.  The real machine has a big advantage... It's done in hardware!  Performance is determined by the workload as in all cases.   More resource sharing causes less scalability.  

RDzJohn

Thanks for an understandable explanation.