Function logical_parallelism 

pub fn logical_parallelism() -> usize

Logical processors this process should parallelize across.

Derived once and cached for the process lifetime. The topology cannot change while the process runs, and deriving it is not cheap: a CpuTopology::detect() call measures 9,935 ns and 77 allocations totalling 16,480 bytes on a 24-processor host, because it materializes the whole NUMA and cache-level description to read one count. Callers that need a worker count per operation must not pay that, so this is the one place the derivation happens.

themis reports the machine’s logical processors; available_parallelism is the fallback when no topology is available.